Adaptive radiation and the topology of large phylogenies.
Key words. - Adaptive radiation, diversification, nuil models, phylogeny, topology.
Evolutionary biologists often suggest that an ancestral species evolved features that allowed entry into a new adaptive zone (Simpson, 1953) and proliferation into the many niches available in this new zone, often to the detriment (extinction) of other organisms. This constitutes one definition of adaptive radiation. For example, the rapid diversification of birds into the aerial environment has been used as a textbook example of adaptive radiation. The advantages of flight are thought to have allowed speciation to occur at a more rapid pace than in the archosaurian relatives of birds. Associated with this proliferation is the notion that the more successful birds outcompeted archosaurian sister groups, resulting in an increased extinction rate among these "reptiles" (Ehrlich et al., 1974).
The arguments generated above to explain bird evolution follow from the observations that 1) there are many living bird species (9,672; Sibley and Monroe, 1990), 2) there are few living archosaurian relatives (21 crocodilian species; Spellerberg, 1982), and 3) these groups share a sister-taxon relationship and, therefore, have had equivalent lengths of time over which speciation and/or extinction have occurred (Cracraft, 1981). One common interpretation of such observations is, by inspection, to deem them unusual and to search for some explanation. Often a deterministic scenario is generated by suggesting that key features (e.g., winged forelimbs, air sacs, and loss of teeth for birds) of one group within a pair of sister taxa confer advantages and consequently are responsible for increased rates of speciation and/or decreased rates of extinction within this lineage. Such observations have been used to explain the diversity of anoline lizards (Peterson, 1983), passerine birds (Fitzpatrick, 1988), rodents (Luckett and Hartenberger, 1985), and flowering plants (Crepet, 1983), among many others.
This traditional approach to identifying adaptive radiations can be criticized because it requires the assumption that the radiation results from nonrandom diversification and that the appropriate key adaptive feature can be identified in an unbiased fashion. The latter assumption call be violated when key features are selected because they are diagnostic of species-rich clades, a procedure that may lead to bias in tests of the association between key features and diversity (Cracraft, 1990). Additionally, because phylogenetic trees must take some shape, species-rich clades can occur that require no deterministic (e.g., adaptive) explanation (Raup et al., 1973). Thus, the traditional approach for identifying adaptive radiations rarely allows an adequate test of adaptive radiation (Cracraft, 1990; Gould and Lewontin, 1979).
A variety of comparative approaches designed to document the nonrandom diversification expected of adaptive radiations has been tried. However, to meet the statistical requirement of independence, evolutionary comparisons must be made on a sister-taxon basis (Felsenstein, 1985). The results of many past studies can be questioned because they compared functional groups (e.g., Levinton, 1974; Herrera, 1989) and/or included paraphyletic taxa (e.g., Stanley et al., 1981; Dial and Marzluff, 1989). Other studies have relied on counts of numbers of units within larger taxa (e.g., species within genera; Flessa and Levinton, 1975; Stanley et al., 1981; Dial and Marzluff, 1989; Herrera, 1989), which may indicate more about how systematists delimit taxa than about evolutionary pattern (Guyer and Slowinski, 1991).
A few studies have improved the analysis of adaptive radiation by providing statistical tests based on sister-taxon comparisons. However, some of these can be faulted for considering only a single sister-taxon comparison (e.g., Vrba, 1984) and, therefore, lack sufficient replication to rule out alternative explanations (Mitter et al., 1988). Also, these studies include a risk of sampling bias because observations used to invoke adaptive radiation may then be used to test for this evolutionary feature. For example, Mitter et al. (1988) tested a hypothesis advanced by Southwood (1973) and refined by Strong et al. (1984), which suggests that phytophagy allows species proliferation in insects. This hypothesis was generated after observing that orders containing phytophagous insects typically contain more species than nonphytophagous orders (Strong et al.. 1984). Mitter et al. (1988) attempted to test this adaptational hypothesis by comparing replicate sister orders, families, and/or subfamilies, one phytophagous and the other nonphytophagous. within the five most species-rich insect orders. They found that phytophagous clades were more species rich than nonphytophagous sister clades in 11 of 13 comparisons. However, five of these sister pairs were described (but not in a phylogenetic context) in an appendix to Strong et al. (1984) and an additional three belong to an order (Hemiptera) already estimated by Strong et al. (1984) to be primarily phytophagous. Thus, the observations used by Strong et al. (1984) in creating the hypothesis were not entirely independent of those used by Mitter et al. (1988) to test it.
The problems indicated above suggest that evolutionary biologists may be disproportionately attracted to groups that appear to represent adaptive radiations while ignoring those that do not. To test adequately whether or not adaptive radiation has been a general feature in evolution, one must develop a prediction that differs from the observation used to generate the hypothesis. In this paper we 1) explore the implications of the concept of adaptive radiation relative to the shapes of phylogenetic trees and 2) build upon our null model approach to test for the presence of evolutionary patterns expected if adaptive radiation has been a general feature of evolution.
Null Models in Phylogenetic Studies
The observations used to invoke adaptive radiation can be reformulated in terms of the topology of phylogenetic trees. If one begins with n labeled (named) taxa, the number of dichotomous topologies can easily be determined by equations given by Wedderburn (1922). These tree topologies span a range of possibilities (Fig. 1). At one extreme are completely unbalanced shapes in which each interior node denotes a dichotomy with one subtree representing a single taxon and the other subtree representing all remaining taxa. At the other extreme are completely balanced topologies in which the basal node of the phylogeny divides the n taxa evenly (two subtrees with r = n/2 and s = r taxa, respectively, when n is an even integer) or as evenly as possible (two subtrees with r = (n - 1)/2 and s = n - r taxa, respectively, when n is an odd integer).
If adaptive radiation is a consistent feature of evolution, then the observations outlined above reduce to the assertion that phylogenetic trees should be consistently unbalanced. Because unbalanced trees are unlikely to result from chance alone, some deterministic explanation is required (e.g., Vermeij, 1988). Quantitative support for these deterministic explanations is difficult to obtain. The question of interest becomes: how likely is it that one could generate, via some appropriate null model, trees as unbalanced as or more unbalanced than ones observed in nature? Note that the test required is one-tailed, because in the alternative hypothesis a deterministic evolutionary process (adaptive radiation) causes more speciation and/or less extinction to occur in one subtree relative to the sister subtree. The advantage of this approach is that an explicit null hypothesis is generated that describes the pattern of tree shapes expected in the absence of adaptive radiation. Failure to reject this hypothesis indicates that no special evolutionary mechanism need be invoked.
A test of the generality of adaptive radiation can be accomplished by considering the frequency with which different tree topologies occur in published phylogenetic reconstructions. This is relatively simple when phylogenies are small because the number of topologies will also be small and, therefore. manageable. Two important models have been used to generate null expectations for a given topology. One null model, promoted by Simberloff et al. (1981), involves allowing growing branch tips of a phylogeny to diverge at random (Raup et al., 1973). By random we mean that each species diverges (speciates) independently of all others and that the probability that a species will diverge at any point in time is equivalent for all species. The expected null frequencies of each topology are then generated by determining the number of paths that lead to each topology. For example, there are six possible paths that an ancestral species, diverging at random, can follow to yield trees with four terminal branches. These trees conform to only two possible topologies, an unbalanced, comb-shaped one and a balanced, bifurcate one (Fig. 2). Four paths lead to the former topology; therefore, its null probability of occurrence is 4/6 = 0.67. Two paths lead to the latter topology; its null probability is 2/6 = 0.33. Following the terminology of Simberloff et al. (1981), we will refer to this as the Markov model. This null model has the advantage of being based upon a speciation process (Savage, 1983) and can be used to determine whether observed topological frequencies differ from those expected due to random speciation and/or extinction events.
In a second null model, implied by Rosen (1978), each possible tree is assumed to be equally likely. The expected null frequencies of each topology are then generated by determining the number of ways to arrange n taxa on each tree topology. In the case of four terminal taxa, there are 15 possible arrangements of these taxa on dichotomous trees (Fig. 3). Twelve of these conform to the unbalanced topology which, therefore. has a null probability of 12/15 = 0.80. Three of the trees conform to the balanced topology, yielding a null probability of 3/15 = 0.20. Again following the terminology of Simberloff et al. (1981), we will refer to this as the proportional-to-distinguishable-arrangements model. This model does not test whether evolution has proceeded in a random fashion. Rather it determines whether the topologies of trees produced by systematists differ, on average, from those expected from random selection of phylogenies from the pool of all possible trees of equal size (Guyer and Slowinski, 1991).
We recommend that the hypotheses be used in tandem. Rejection of the proportional-to-distinguishable-arrangements hypothesis provides evidence that trees are not random guesses at the true phylogeny; failure to reject this hypothesis indicates either that evolution favors tree shapes predicted by this model or that systematic efforts are not sufficiently well corroborated to be distinguished from random guesses. Rejection of the Markov null model suggests that some deterministic process shapes evolutionary trees; failure to reject this hypothesis indicates that deterministic processes need not be invoked (Guyer and Slowinski, 1991).
For phylogenies with many taxa, the number of possible topologies quickly becomes unwieldy, necessitating a different analytical approach. We have focused on the basal node (root) in assessing null models for large phylogenies. This node partitions the n taxa into subtrees representing r and s taxa, where r [is less than or equal to] s. If systematists' estimates of phylogenetic histories for large radiations differ from random guesses then the disparity in the number of taxa found in the two subtrees should differ from the proportional-to-distinguishable-arrangements model. If adaptive radiations are a frequent feature of evolution, then one would expect r to be significantly less than s when averaged over several phylogenies. This disparity should be greater than that expected from the Markov model. Elsewhere we have derived equations to determine, for any n-member tree, the probability of observing a dichotomy at the basal node resulting in groups with r and s taxa under the proportional-to-distinguishable-arrangements and the Markov null models (Slowinski, 1990; Slowinski and Guyer, 1989).
For the example of bird evolution, we know that some ancestor diverged into two sister lineages, one representing the 21 living crocodilian species and the other including the 9,672 living bird species. Thus, the appropriate sampling universe is the pool of all possible monophyletic lineages comprising 9,693 total taxa. In theory, it would have been possible to sample, at random, a single tree from this universe and to have drawn the bird-crocodile tree. The topology of this tree could then be tested against the two null hypotheses. In practice, such random sampling is not accomplished. In fact, the feature of bird evolution that has attracted so much attention is precisely the fact that birds appear to be impressively species rich. Therefore, one must question how many unimpressive radiations of vertebrates biologists overlooked before attempting to explain the pattern of diversity seen in birds. This sampling bias is an underappreciated problem with adaptive radiation scenarios. The question of whether large phylogenies as a general rule exhibit a non-Markovian pattern has yet to be tested and is the focus of the following analysis.
A Test of the Generality of
We sampled published large phylogenies in an attempt to evaluate whether adaptive radiation has occurred frequently enough to cause the predicted effect on the shape of phylogenetic trees indicated above. Our goal in this search was to answer two questions: 1) Do tree shapes differ among major phylogenetic groups (i.e., has large-scale evolution proceeded differently among monophyletic higher taxa)? 2) Are large phylogenies ([is greater than or equal to] 100 species) non-Markovian relative to the diversity between pairs of sister taxa (i.e., are evolutionary patterns of large trees different from the pattern generated by random divergence of growing branch tips)?
Ten large trees were collected from each of three major lineages; division Angiospermae, class Insecta, and subclass Tetrapoda. To be included in our survey, the trees were subjected to the following constraints: 1) all phylogenies must have been constructed with character data for which the data matrix was presented or for which character state changes were mapped on a tree, 2) the ingroup must have consisted of at least 100 species, 3) the basal node for the ingroup was selected as the point of interest within the phylogeny and must have been dichotomous with character state changes indicated for the three branches incident on this node. This was done to insure that monophyly was established for each subtree and that the two subtrees arising from the root were sister taxa. In most cases the basal node refers to the actual basal node of the published phylogeny. However, when the species diversity of some taxon considered in the ingroup was difficult to ascertain, we selected another interior node within the published phylogeny. We refer to these interior nodes as basal nodes, because that reflects our use of them in the analyses. All node selections used in this study are identified in the Appendix. We did not try to assess the quality of the trees by critiquing the data used to generate them or demanding that a minimum level of consistency of the data with the published trees be reached. Instead, we sampled trees that were representative of large phylogenies in the current systematic literature. The basal node of the ingroup was used to assess whether the species richness of the two subtrees arising from it differed from the proportional-to-distinguishable-arrangements and/or Markov models. When not provided by the primary article, numbers of species in these two subtrees were estimated by species counts given in general texts or literature (see Appendix). We chose not to include the node attached to the outgroup because outgroups may be selected because they are small and manageable, rather than because they are the closest living relatives of the ingroup. Because the trees were selected without regard to their shape, our sample should be unbiased and provides a first step in documenting the presence of evolutionary patterns expected of adaptive radiation. Additionally, because comparative studies of adaptive radiation should involve examination of sister taxa that differ in the presence or absence of some key feature (Felsenstein, 1985) we believe that our single-node approach correctly reflects comparisons made in current studies of adaptive radiation (e.g., Mitter et al., 1988). However, we examine below the effect of selecting nodes above the basal node.
The 30 sampled phylogenies were distributed among many orders within the three major groups and great variation occurred in the degree of unbalance (proportion of taxa found in the more diverse group) observed in the trees (Table 1). This was true of all three radiations as proportions ranged from 0.505 to 0.999. Thus, as one might expect, large phylogenies can be quite unbalanced, quite balanced, and all conditions between these extremes.
We used data regarding the distribution of taxa along subtrees at the basal node (Table 1) to ask the question, are the distributions of taxa different among taxonomic groups? We found no significant difference in the allocation of diversity at the basal node among the three taxonomic radiations (one-way ANOVA for arcsine transformed proportions; F = 2.37, P = 0.112). Thus, whatever the process(es) of evolution involved in causing divergences within angiosperms, insects, and tetrapods, a similar pattern of diversity occurs along the branches of the basal dichotomy. We described similar results for a sample of small phylogenies (Guyer and Slowinski, 1991).
To assess whether observed large trees differ from proportional-to-distinguishable-arrangements and Markov predictions, we defined unbalanced trees as being those that had [is greater than or equal to] 90% of the taxa along the more diverse branch of the basal dichotomy. This arbitrary cutoff was chosen because it 1) delimits a group of phylogenies in which most speciation has occuffed along one basal branch and not its sister-taxon branch and 2) provides expected cell frequencies of acceptable size for contingency table analyses. Our sample of large trees contains 15 unbalanced and 15 balanced examples. The proportional-to-distinguishable-arrangements hypothesis predicts 25.5 unbalanced and 4.5 balanced trees whereas under the Markov model one would expect to observe 6 unbalanced and 24 balanced trees (expected values determined from equations given by Slowinski, 1990 and Slowinski and Guyer, 1989). Thus, published large phylogenies differ both from the proportional-to-distinguishable-arrangements ([X.sub.2] = 28.8, P < 0.001) and Markov ([X.sub.2] = 16.9, P < 0.001) models.
Rejection of the proportional-to-distinguishable-arrangements hypothesis indicates that systematists' attempts to reconstruct the phylogenetic history of large radiations of organisms can be differentiated from random selection of a history for each group. This conclusion differs from a similar analysis of topological frequencies of published small phylogenies five terminal taxa; Guyer and Slowinski, 1991). The small trees, sampled from the systematic literature without regard to the strength of the data, conform to topological frequencies expected if systematists guessed at phylogenies. We concluded that most small trees were constructed from too few characters to adequately diagnose interior nodes. The results of the present study indicate that at least the basal nodes of large phylogenies are sufficiently well delimited to differentiate attempts at phylogenetic reconstruction from one null hypothesis of random tree selection. Additionally, the number of character state transitions occurring at the root and the two branches attached to the root for our sample of large trees (x = 6.2, SE = 1.3, N = 90) was greater than the character support for interior nodes in our sample of small trees (x = 2.5, SE = 0.1, N = 300). This finding indicates that our sample is adequate for assessing the Markov model. The results for the Markov model suggest that evolution of large phylogenetic groups did not result from random divergence of tips of a growing tree and that, on average, large trees do tend to be significantly unbalanced. This finding justifies a search for some deterministic cause for nonrandom patterns of tree shape and is consistent with the hypothesis that adaptive radiation has been a general feature of evolution.
The finding that published phylogenies are not as unbalanced as predicted by the proportional-to-distinguishable-arrangements model might seem incongruous, given the prediction that trees should be unbalanced if adaptive radiation is a general phenomenon. This result merely illustrates the utility of null models in evolutionary biology. If systematists selected a tree at random from the pool of all possible trees, most of them would conform to the completely unbalanced topology (e.g., Fig. la) when examined at the basal node (Slowinski, 1990). This is because there are many more ways to arrange labeled objects on this topology than on any other. Thus, unbalanced trees are expected if adaptive radiations occur often or if systematists perform no better than guessing at the true phylogeny. To demonstrate adaptive radiation, trees must be more unbalanced than those expected by the proportional-to-distinguishable-arrangements null model or less unbalanced than this model, but more unbalanced than the Markov model (the observation made for all three groups in this study).
Thus far, all of our analyses have dealt with the basal node. We can envision one way that these nodes might be biased. The taxonomic placement of some groups has remained problematic despite the best efforts of systematists. For example, if the family Curculionidae had been selected as a potential example of adaptive radiation among insects, then one would have to determine whether to include platypodids and scolytids as subfamilies within the Curculionidae (as suggested by Crowson, 1967) or whether to exclude them as distantly related groups outside Curculionidae (Wood, 1986). If our collection of published trees consistently included such problematic lineages then we would expect them to be placed most frequently at the basal node. This would occur because such lineages would share few apomorphies with the other ingroup lineages (otherwise there would be no problematic groups). If problematic groups tended to have reduced species richness then unbalanced phylogenies would be expected at the basal nodes but not at nodes occurring higher in the trees. To test this we determined the proportions characterizing the basal six nodes of three lineages within each of flowering plants, insects, and tetrapods (Table 2). A nested, repeated-measures ANOVA was used to test for a significant difference among nodes (repeated measures) within lineages (nested within flowering plants, insects, and tetrapods). No significant node effect was found within our sampled trees (F = 1.73, P = 0. 16), indicating that our results are not biased by our use of a single basal node.
Summary and Conclusions
One definition of adaptive radiation is that some organisms have features that allow them to speciate more prolifically or become extinct less frequently than organisms without these features. If this type of adaptive radiation was a consistent feature of evolution then large phylogenetic trees should be significantly unbalanced. Despite recent advances in the comparative method in evolutionary biology, this prediction for topologies of large trees has remained inadequately tested. Instead, testing has proceeded in a fashion that may involve sampling bias because attention may be placed on impressively unbalanced trees while more balanced trees are overlooked.
Our test of topologies focused on the basal nodes of large phylogenies. We used two null models to test 1) whether published trees differed in shape from those expected from random selection of trees from the pool of all possible similarly sized phylogenies and 2) whether they were more unbalanced than expected due to random speciation and/ or extinction. Our results indicate that the factors affecting tree shape have acted consistently among tetrapods, insects, and flowering plants. The degree of unbalance in published large phylogenies cannot be explained by weak systematic treatment because our results indicate a nonrandom selection of trees by systematists. On average, speciation has been favored and/or extinction reduced along certain branches within large radiations, justifying the tendency to attribute such radiations to some deterministic cause. This represents an important first step in establishing the presence of a pattern expected if adaptive radiation has been a general phenomenon in evolution.
Our analyses recognize adaptive radiations only when one sister group develops increased speciation and/or decreased extinction rates while the other sister lineage remains unchanged. Other types of adaptive radiations might be envisioned. For example, both sister lineages might evolve features that alter speciation and/or extinction rates more or less simultaneously. In this case a balanced basal node would be expected. We viewed such balanced radiations as being less likely than unbalanced ones because key features would have to appear independently in each sister taxon. For this reason we maintained a restricted definition of adaptive radiation. However, the null model approach could be modified to assess less restricted definitions. Care must be taken, however, to insure that the concept does not become so broadly defined that it loses its ability to be tested against some null hypothesis.
Because the samples were taken without regard to the partitions represented by the basal node, our results improve previous attempts to demonstrate nonrandom patterns of diversification and provide hope that carefully designed comparative studies (e.g., Farrell et al., 1991; Mitter et al., 1988; Zeh et al., 1989) can correctly associate putative adaptive features with nonrandom patterns of diversification. This conclusion must be tempered, however, with 1) the caution offered by others (e.g., Cracraft, 1990; Gould and Calloway, 1980; Mitter et al., 1988) against uncritical use of adaptive radiation scenarios and 2) the acknowledgment that the comparative method rests on phylogenetic soil that may shift dramatically as data accumulate (e.g., Gauthier et al., 1988).
Cracraft, J. 1981. Pattern and process in paleobiology: The role of cladistic analysis in systematic paleontology. Paleobiology 7:456-468. _____. 1990. The origin of evolutionary novelties: Pattern and process at different hierarchical levels, pp. 21-44. In M. H. Nitecki (ed.), Evolutionary Novelties. University of Chicago Press, IL USA. Crepet, W. L. 1983. The role of insect pollination in the evolution of the angiosperms, pp. 29-50. In L. Real (ed.), Pollination Biology. Academic Press, Orlando, FL USA. Crowson, R. A. 1967. The natural classification of
the families of Coleoptera. Classey, Hampton, UK. Dial, K. P., and J. M. Marzluff. 1989. Nonrandom diversification within taxonomic assemblages. Syst. Zool. 38:26-37. Ehrlich, P. R., R. W. Holm, and D. R. Parnell. 1974. The Process of Evolution. McGraw-Hill, San Francisco, CA USA. Farrell, B.D., D. Dussourd, and C. Mitter. 1991. Escalation of plant defense: Do latex/resin canals spur plant diversification? Am. Nat. 138:881-900. Felsenstein, J. 1985. Phylogenetics and the comparative method. Am. Nat. 125:1-15. Fitzpatrick, J. W. 1988. Why so many passerine birds? A response to Raikow. Syst. Zool. 37:71-76. Flessa, K. W., and J. S. Levinton. 1975. Phanerozoic diversity patterns: Tests for randomness. J. Geol. 83:239-248. Gauthier, J., A. G. Kluge, Am T. Rowe. 1988. Amniote phylogeny and the importance of fossils. Cladistics 4:105-209. Gould, S. J., and C. B. Calloway. 1980. Clams and brachiopods - Ships that pass in the night. Paleobiology 6:383-396. Gould, S. J., and R. C. Lewontin. 1979. The spandrels of San Marco and the Panglossian paradigm: A critique of the adaptationist programme. Proc. R. Soc. London 205:581-598. Guyer, C., and J. B. Slowinski. 1991. Comparisons of observed phylogenetic topologies with null expectations among three monophyletic lineages. Evolution 45:340-350. Herrera, C. M. 1989. Seed dispersal by animals: A role in angiosperm diversification. Am. Nat. 133: 309-322. Levinton, J. S. 1974. Trophic group and evolution in bivalve molluscs. Palaeontology 17:579-585. Luckett, W. P., AND J.-L. Hartenberger. 1985. Preface, pp. vii-ix. In W. P. Luckett and J.-L. Hartenberger (eds.), Evolutionary Relationships among Rodents. Plenum Press, N.Y., USA. Mitter, C., B. Farrell, and B. Wiegmann. 1988. The phylogenetic study of adaptive zones: Has phytophagy promoted insect diversification? Am. Nat. 132:107-128. Peterson, J. A. 1983. The evolution of the subdigital pad in Anolis. I. Comparisons among the anoline genera, pp. 245-283. In G. J. Rhodin and K. Miyata (eds.), Advances in Herpetology and Evolutionary Biology. Harvard University, Cambridge, MA USA. Raup, D. M., S. J. Gould, T. J. M. Schopf, and D. S. Simberloff. 1973. Stochastic models of phylogeny and the evolution of diversity. J. Geol. 81: 525-542. Rosen, D. E. 1978. Vicariant pattern and historical explanation in biogeography. Syst. Zool. 27:159-188. Savage, H. M. 1983. The shape of evolution: Systematic tree topology. Biol. J. Linn. Soc. 22:225-244. Sibley, C. G., and B. L. Monroe. 1990. Distribution and taxonomy of birds of the world. Yale University Press, New Haven, CT USA. Simberloff, D., K. L. Hecht, E. D. McCoy, and E. F. Connor. 1981. There have been no statistical tests of cladistic biogeographical hypotheses, pp. 40-63. In G. Nelson and D. E. Rosen eds,), Vicariance Biogeography: A Critique. Columbia University Press, N.Y., USA. Simpson, G. G. 1953. The Major Features of Evolution. Columbia University Press, N.Y., USA. Slowinski, J. B. 1990. Probabilities of n-trees under two models: A demonstration that asymmetrical interior nodes are not improbable. Syst. Zool. 39: 89-94. Slowinski, J. B., and C. Guyer. 1989. Testing the stochasticity of patterns of organismal diversity: An improved null model. Am. Nat. 134:904-921. Southwood, T. R. E. 1973. The insect/plant relationship - An evolutionary perspective. Symp. R. Entomol. Soc. London 6:3-30. Spellerberg, I. F. 1982. Biology of Reptiles: An Ecological Approach. Chapman and Hall, N.Y., USA. Stanley, S. M., P. W. Signor III, S. Lidgard, and A. F. Karr. 1981. Natural clades differ from "random" clades: Simulations and analyses. Paleobiolog), 7:115-127. Strong, D. R., J. H. Lawton, and T. R. E. Southwood. 1984. Insects on Plants: Community Patterns and Mechanisms. Harvard University Press, Cambridge, MA USA. Vermeij, G. J. 1988. The evolutionary success of passerines: A question of semantics? Syst. Zool. 37: 69-71. Vrba, E. S. 1984. Evolutionary pattern and process in the sister-group Alcelaphini-Aepycerotini (Mammalia: Bovidae), pp. 62-79. In N. Elderidge and S. M. Stanley (eds.), Living Fossils. Springer-Verlag, N.Y., USA. Wedderburn, J. H. M. 1922. The functional equation g( [x.sub.2] ) = 2 [alpha] x + [[g(x)].sup.2]. Ann. Math. (2nd Ser.) 24:121-140. Wood, S.L. 1986. Are classification of the genera of Scolytidae (Coleoptera). Great Basin Nat. Mem. 10: 1-126. Zeh, D. W., J. A. Zeh, and R. L. Smith. 1989. Ovipositors, amnions and eggshell architecture in the diversification of terrestrial arthropods. Q. Rev. Biol. 64:147-168.
The references listed here are the sources used in this study for large phylogcnies. Numbers given before each citation correspond to those given in Tables 1 and 2. Figure number or page number is given for each source. Parenthetical notations indicate that numbers = number of large clades used from the published tree; N = study used in analysis of node effect (Table 2). Unnumbered references were used as sources for species richness in each terminal taxon. See text for explanation.
1. Burns-Balogh, P., and V. A. Funk. 1986. A phylogenetic analysis of the Orchidaceae. Smithson. Contrib. Bot. no. 61. Figure 1 starting at basal node (1,N), Gastrodieae + Triphoreae assumed to be monophyletic. 2. Burns-Balogh, P., and H. Robinson. 1983. Evolution and phylogeny of the Petexia alliance (Orchidaceae: Spiranthoideae: Spiranthinae). Syst. Bot. 263-268. Figure 1 starting with node labeled EFG, (1). 3. Campbell, C. S., and E. A. Kellogg,. 1986. Sister group relationships of the Poaceae. In T. R. Sodderstrom, K. W. Hilu, C. S. Campbell, and M. E. Burkworth (eds.), Grass Systematics and Evolution. Smithsonian Institute Press, Washington, DC USA. Figure 20.1 starting at node linking Centrolepidaceae with all other Poales, (1). Number of species from Heywood, 1978. 4. Crisp, M. D., and P. H. Weston. 1987. Cladistics and legume systematics, with an analysis of the Bossiaeeae, Brongniartieae and Mirbelieae (Papilionoidea, Leguminosae). ln C. H. Stirton (ed.), Advances in Legume Systematics, Part 3. Royal Botanic Gardens, Kew, UK. Figure 10a starting at node uniting Bossiaeeae and Mirbelieae, (1). Number of species in Bossiaeeae from Hutchinson, 1964. 5. Goldblatt, P. 1990. Phylogeny and classification of Iridaceae. Ann. Mo. Bot. Gard. 77:607-627. Figure 3 starting at basal node, (1). 6. Jansen, R. K., H. T. Michaels, J. D. Palmer. 1991. Phylogeny and character evolution in the Asteraceae based on chloroplast DNA restriction site mapping. Syst. Bot. 16:98-114. Figure 1 starting at node labeled 12, (1). 7. Johnson, L. A. S., and B. G. Briggs. 1984. Myrtales and Myrtaceae - A phylogenetic analysis. Ann. Mo. Bot. Gard. 71:700-756. Figure 7 starting at node uniting TRA and ONA and node uniting STH and CMB+LAG, (2). Number of species from Heywood, 1978 and Hutchinson, 1964. 8. Kress, W. J. 1990. The phylogeny and classification of Zingiberales. Ann. Mo. Bot. Gard. 77:698-721. Figure 7 starting at basal node, (1,N). 9. Lavin, J. D. 1991. Tribal relationships of Sphinctospermum (Leguminosae): Integration of traditional and chloroplast DNA data. Syst. Bot. 16:162-172. Figure 1 starting at node uniting all Robinieae, (1). 10. Phipps, J. B.. K. R. Robertson, J. R. Rohrer, and P. G. Smith. 1991. Origins and evolution of Subfam. Maloideae (Rosaceae). Syst. Bot. 16:303-332. Figure 11 starting at node uniting Eriolobus and clade containing Mespilus (1,N). Number of species from Robertson et al., 1991. Heywood, V. H. 1978. Flowering Plants of the World. Oxford University Press, Oxford, UK. Hutchinson, J. 1964. The Genera of Flowering Plants. Oxford University Press, Oxford, UK. Robertson, K. R., J. B. Phipps, J. R. Rohrer, and P. G. Smith. 1991. A synopsis of genera in Maloideae (Rosaceae). Syst. Bot. 16:376-394.
1. Askevold, J. S. 1990. Reconstructed phylogeny and reclassification of the genera of Donaciinae (Coleoptera: Chrysomelidae). Quaest. Ent. 26:601-664. Figure 15 starting at basal node. 2. Harrington, B. J. 1980. A generic level revision and cladistic analysis of the Myodochini of the world (Hemiptera, Lygaeidae, Rhyparochrominae). Bull. Am. Mus. Nat. Hist. 167:45-116. Figure 103 starting at node 1, (1). 3. Herman, L. H. 1986. Revision of Bledius. Part IV. Classification of species groups, phylogeny, natural history, and catalogue (Coleoptera, Staphylinidae, Oxytelinae). Bull. Am. Mus. Nat. Hist. 184:1-368. Figure 41 starting at node labeled Bledius, (1,N). 4. Kimsey, L. S. 1987. Generic relationships within the Euglossini (Hymenoptera: Apidae). Syst. Entomol. 12: 63-72. Figure 29 starting at basal node, (1). 5. Lyal, C. H. C. 1985. Phylogeny and classification of the Psocodea, with particular reference to the lice (Psocodea: Phthiraptera). Syst. Entomol. 10:145-165. Figure 2 starting at basal node, (1). 6. Miller, J. S. 1987. Phylogenetic studies in the Papilioninae (Lepidoptera: Papilionidae). Bull. Am. Mus. Nat. Hist. 186:365-512. Figures 3, 8, and 15; 211 tarting at basal node, [3,N(fig. 3)]. 7. Morse, J. C., and R. W. Holzenthal. 1985. Higher classification of Triplectidinae (Trichoptera: Leptoceridae). In M. Bournaud and H. Tachet (eds.), Proceedings of the 5th International Symposium on Trichoptera, Lyon, France, 21-26 July 1986. Junk Publ., Dordrecht, Netherlands. Figure 1 starting with node labeled Triplectidinae, (1). 8. Williams, P. H. 1985. A preliminary cladistic investigation of relationships among bumble bees (Hymenoptera: Apidae). Syst. Entomol. 10:239-255. Figure 9, starting at node labeled BOMBUS, (1,N).
1. Cracrafr, J. 1985. Monophyly and phylogenetic relationships of the Pelecaniformes: A numerical cladistic analysis. Auk 102:834-853. Figure 7 starting at node 4, (1). Number of species from Welty and Baptista, 1988. 2. Duellman, W. E., and L. Trueb. 1986. Biology of Amphibians. McGraw-Hill, N.Y. USA. Figure 17-1 starting at node linking Sirenidae with other salamanders, Figure 17-3 starting at basal node, [2,N(fig. 17-1)]. 3. Frost, D. R., and R, Etheridge. 1989. a Phylogenetic analysis and taxonomy of iguanian lizards (Reptilia: Squamata). Misc. Publ. Mus. Nat. Hist., University of Kansas, no. 81. Figure 13 starting at basal node, (1). 4. Gaffney, E. S., and P. A. Meylan. 1988. A phylogeny of turtles. In M. J. Benton (ed.), The Phylogeny and Classification of Tetrapods, Vol. 1: Amphibians, Reptiles, Birds. Clarendon Press, Oxford, UK. Figure 5.1 starting at node A5. Number of species from Pritchard, 1979. 5. Guyer, C., and J. M. Savage. 1986. Cladistic relationships among anoles (Sauria: Iguanidae). Syst. Zool. 35:509-53 1. Figure 5 starting at node linking Chamaeleolis with other anoles, (1). 6. Kluge, A. G. 1987. Cladistic relationships among the Gekkonoidea. Misc. Publ. Mus. Zool., University of Michigan, no. 173. Figure 11 starting at node 1, (1). 7. Miyamoto, M. M., and M. Goodman. 1986. Biochemical systematics of eutherian mammals: Phylogenetic patterns and classification. Syst. Zool. 35:230-240. Figure 3 starting at basal node, (1,N). Number of species from Vaughan, 1986. 8. Presch, W. F. 1988. Cladistic relationships within the Scincomorpha. In R. Estes, G. Pregill (eds.), Phylogenetic Relationships of Lizard Families. Stanford Press, Palo Alto, CA USA. Figure 5, (1,N). 9. Swierczewski, E. V., and R. J. Raikow. 1981. Hind limb morphology, phylogeny and classification of the Piciformes. Auk 98:466-480. Figure 1, (1). Number of species from Welty and Baptista, 1988. Pritchard, C. H. 1979. Encyclopedia of Turtles. T. F. H. Publications, Neptune, NJ USA. Vaughan, T. A. 1986. Mammalogy. Saunders, Philadelphia, PA USA. Welty, J. C., and L. Baptista. 1988. The Life of Birds, Saunders, San Francisco, CA USA.
|Printer friendly Cite/link Email Feedback|
|Author:||Guyer, Craig; Slowinski, Joseph B.|
|Date:||Feb 1, 1993|
|Previous Article:||Reversibility in evolution: a maximum likelihood approach to character gain/loss bias in phylogenies.|
|Next Article:||Isolation by distance in equilibrium and non-equilibrium populations.|