Phylogenetic tests of speciation hypotheses

4Phylogenetic Patterns as Tests

of Speciation Models

Peter J. Wagner and Douglas H. Erwin

Most-paleobiological discussions about phylogeny have focused on par-ticular transitions from an ancestral species to a descendant one, or on afew transitions within a single taxon (e.g., Gingerich 1985). Althoughvaluable, these studies represent individual examples of a pattern of spe-ciation rather than tests of the generality of any particular mode. More-over, adequately documented speciation events are very rare. Conse-quently, even if the investigators have a hypothesis about a clade'sphylogeny, they usually lack any direct evidence of the evolutionary pro-cesses that produced that phylogeny.

A different approach to testing hypotheses about patterns of speciationamong fossil species involves a strategy of examining phylogenetic pat-terns. Different speciation models predict different relationships betweenspeciation and the survival of ancestral morphologies. More specific mod-els also postulate different associations between species-level traits (e.g.,temporal and geographic ranges) and the likelihoods of leaving de~cen-dants ..Therefore, several additional predictions stem from the hypothesis

j that a particular mode of speciation (vicariance, peripheral isolation, etc.)predominates within a clade. These include which patterns of speciation(anagenesis, bifurcation, cladogenesis) and phylogenetic branching (po-lytomies, pectinate branches, etc.) should be most common, and the rel-ative likelihoods of sampling plesiomorphic and apomorphic species fromthe fossil record.

88 Peter J. Wagner and Douglas H. Erwin

Because hypotheses about modes of speciation ultimately make pre-dictions about the phylogenetic geometries of clades, phylogenetic pat-terns can be used to corroborate or refute hypotheses predicting that par-ticular modes of speciation were dominant among closely related taxa.Here we will provide examples of the phylogenetic patterns produced by-cladogenesis, anagenesis, and bifurcation. Although reticulation certainlyis common among plants and possibly more common among animalsthan generally realized (e.g., Smith 1992), it is much more difficult torecognize in the fossil rec?rd. We therefore will not discuss examples ofreticula tion.

We will use two previously published computer simulations of phy-logeny to highlight how different assumptions about speciation patternsproduce very different phylogenetic patterns. We then use some basicassumptions derived from the simulated results to examine two examplesfrom the fossil record. We aJso discuss why cladograms alone are insuf-ficient to describe phylogenetic patterns without hypothesizing aboutancestor-descenda,nt relationships and evaluating temporal or geographicdata. Finally, we examine why phylogenies must be interpreted withoutpreconceptions about how speciation proceeds and the implications ofthis for phylogenetic systematics and comparative biology.

Definitions of TermsBecause patterns of speciation have been defined differently by variousauthors, we will define exactly how we are using each term. The originaldefinition of "dadogenesis" (Rensch 1959, p. 97) included two patterns:(1) ancestral lineages giving rise to two daughter lineages, with the an-cestral morphology disappearing, and (2) ancestral lineages "branching"into new daughter lineages with no necessary change co-occurring in theancestral morphology (see figure 4.1). Some modern definitions (e.g.,Maddison and Slatkin 1991; Nixon and Wheeler 1992) restrict clado-genesis to the first pattern. Other definitions (e.g., Raup 1977; Raup andGould 1974) restrict cladogenesis to the second.

The definition that restricts cladogenesis to the first pattern incorpo-rates elements of anagenesis (i.e., the replacement of an ancestral mor-phology by a derived morphology within a single lineage) because it as-sumes that the ancestral morphology becomes "pseudoextinct."Combining anagenesis and cladogenesis is confusing; therefore we restrict

Phylogenetic Patterns as Tests of Speciation Models 89

our own definition of cladogenesis to the second pattern and use "bifur-cation" to describe the first pattern. In addition, although some workersdo not consider speciation to occur via anagenetic evolution ("phyleticevolution" versus "speciation," sensu Larson 1989), we will consider bi-furcation a pattern of speciation here.

Table 4.1 lists some modes of speciation and the patterns of speciationand phylogenetic topologies that they predict. As with patterns of speci-ation, the definitions of modes of speciation have varied. Our "vicari-

a b

Figure 4.1 The two patterns encompassed by Rensch's (1959) definition of"cladogenesis." (a) Ancestral species become pseudo-extinct (t) during a bifur-cation into two derived daughter lineages. (b) Derived species branch off fromancestral ones, with the ancestral species conserving a nonderived morphology.Here we use cladogenesis to refer to the second pattern only, while we definethe first pattern as bifurcation.

Table 4.1Modes and Patterns of Speciation

Mode Speciation Pattern (typical) Phylogenetic Pattern

Selection-Driven DivergenceSympatryParapatryVicarianceAnagenesisPeripheral IsolationHybridization

BifurcationBifurcationBifurcationBifurcationPseudo-extinctionCladogenesisReticulation

SymmetricSymmetricSymmetricSymmetricPectinatePolytomy?

SOURCES: "symmetric" = "balanced" of Heard 1992; "pectinate" = "Hennigian comb" of Panchen1992 and "unbalanced" of Heard 1992.

NOTES: Hypothesized modes of speciation are matched with the typical speciation patterns eachpredicts and with the phylogenetic pattern expected if that mode predominates within a clade.

CharactersSpecies I II III IV V A B C D E

A 1 0 0 0 0B 1 1 0 0 0C 1 1 1 0 0D 1 1 0 1 0E 1 1 0 0 1

a


ance" encompasses both the "type II" and "type III" allopatry of Brooksand McLennan (1991). We use "peripheral isolation" (sensu Mayr 1963= "peripatry" of Mayr 1982) for Brooks and McLennan's "type I" al-lopatry. Because definitions of allopatry have encompassed very differentpatterns of speciation, we do not use that term here.

Effects of a Priori Assumptions of InterpretingPhylogenetic PatternIf we are to use phylogenies to test hypotheses about speciation patterns,then the methods of phylogenetic reconstruction must not assume a par-ticular speciation model. Many workers appear to think that cladistic(parsimony) analyses require a bifurcating pattern of speciation (e.g., Lor-enzen and Sieg 1991). This is not entirely correct. For example,the matrixshown in figure 4.2 yields a single most parsimonious tree with a poly-tomy. Nevertheless, many phylogenetic systematists do assume that bi-furcation is the standard pattern of speciation (e.g., Hennig 1966; Mad-dison 1989; Slowinski and Guyer 1989a,b; Nixon and Wheeler 1992;Heard 1992) when interpreting cladograms. Such systematists would notinterpret the polytomy in figure 4.2 as one species giving rise to four butas an artifact of ignorance. The assumption (often explicit) is that most,if not all, polytomies are the result of inadequate data, and that inclusionof more characters would "resolve" some bifurcating pattern.

The assumption of bifurcation obviously imposes a particular patternof speciation on the interpretation of evolutionary patterns and ignoresspeciation models that allow a single species to produce multiple daughter

Figure 4.2 A data matrix (a) whose most parsimonious solution produces apolytomy, shown in (b). Without other data, it cannot be determined whetherthis pattern is due to a single species giving rise to several others or to insuffi-cient data. However, this topology should be found by any parsimony algo-rithm that does not assume bifurcation.


species. To avoid such assumptions, we will take a cue from recent meth-odological works in comparative biology (e.g., Harvey and Pagel 1991)and assume that the cladograms presented here are more or less correcteven if they illustrate polytomies. We therefore accept some polytomiesas signal rather than noise.

Models of Speciation and the Phylogenetic PatternsThey PredictFigure 4.3 shows the phylogeny of a bifurcating lineage. Commonly citedmechanisms for bifurcation include selection-driven divergence (e.g., Dar-win 1859; Gingerich 1976), sympatry (e.g., Maynard Smith 1966; Wake,Yanev, and Frelow 1989), parapatry (e.g., Lande 1982), and vicariance(e.g., Lynch 1989; Brooks and McLennan 1991). Although it has beennoted that a descendant lineage could retain a plesiomorphic morphology(Nixon and Wheeler 1992), this is not the prediction of selection-drivendivergence nor is it a likely prediction of vicariance (e.g., see Brooks andMcLennan 1991).

Slowinski and Guyer (1989a,b; see also Guyer and Slowinski 1991)used three algorithms to generate cladograms, one of which generatedphylogenies using only bifurcation. These simulations assumed that allspecies were equally likely to leave descendants, so extrinsic features suchas temporal and geographic raJlges were not relevant to the outcomes.Also, the possibility of true extinction was ignored. If all species are as-sumed to be extant and only bifurcation occurred, only three evolutionarytrees are possible (figure 4.4a-c). Each of these possible trees can be de-picted as a cladogram (d from a, e from b, f from c). These cladogramseach assume that the ancestral species became "pseudoextinct" after eachspeciation event. If, however, these assumptions are relaxed and fossiltaxa and polychotomies are allowed, then the addition of ancestral speciesproduces very different cladistic patterns for the same original trees (cla-dograms g-i).

If one predicts (or merely assumes) that bifurcating patterns of speci-ation predominated within a clade, then one also is hypothesizing thatthe apparent extinction of plesiomorphic species should be statisticallyindistinguishable from the apparent originations of daughter lineages.(For discussions on statistical tests of the apparent origins and extinctionsof fossil species, see Marshall 1990 and this volume.) Bifurcation modelsalso predict that two derived sister species should share statistically in-

Figure 4.3 A bifurcating phylogenetic pattern. Although this pattern some-times is terme4 "cladogenesis," the pattern shown here is equally anagenetic, asancestral morphologies are replaced by descendant morphologies. Therefore,we have reserved the term cladogenesis for a different phylogenetic pattern. Inthis and in all following figures, ancestral species are given in gray whereasthose without descendants are given in black.

Figure 4.4 Three possible phylogenies (a-c) for five species if bifurcation isthe only mode of speciation. Note that we assume no extinction and completesampling of the taxon. "t" denotes the pseudo-extinction of the ancestral line-age at each bifurcation. (d-f) Three cladograms for the final five species pro-duced in Slowinski's and Guyer's (1989) simulations. These simulate samplingspecies from one time line (e.g., the present), given the assumption that ances-tral morphologies cannot be sampled. (g-j) The previous three cladograms withancestral species included.

I_II I


~l-l'" --1a

Figure 4.5 Phylogeny of anagenetic change (a) with corresponding cladograrn(b). Although each "species" is represented by a vertical band in (a), this ismeant solely to represent the temporal range of a morphotype, not to implymorphologic stasis ..

distinguishable first appearances in the fossil record. Although someworkers have considered the latter statement to be true of all sister taxa(e.g., Cracraft 1981), it is necessarily true only for cases in which speci-ation actually followed a bifurcating pattern.

Figure 4.Sa depicts a lineage in which only anagenesis occurs. A cla-dogram of "morpho species" from this lineage should be completely pec-tinate (sensu Slowinski and Guyer 1989a,b; = "unbalanced" of Heard1992) and all of the lineages save for one would be plesiomorphic (figure4.Sb). Both gradual and punctuated models can predict anagenetic spe-ciation (Wright 1931, 1932; see Jablonski 1986, Provine 1989, andMacLeod 1991). Boucot (1978) has argued that this is indeed the domi-nant pattern observed in the fossil record. Predictions that anageneticmodes of speciation predominated within a clade thus posit that plesio-morphic species should be fairly common and that the temporal rangesof those species should cease where the ranges of (relatively) apomorphicspecies begin. One might also predict general congruence between thegeographic ranges of plesiomorphic and apomorphic species.

Speciation models that predict cladogenesis include models of periph-eral isolation (e.g., Mayr 1963; Eldredge 1971; Eldredge and Gould 1972)and some interpretations of the shifting balance theory (e.g., Wright 1982;Eldredge, this volume). These models do not predict that the evolution of

. a daughter species necessarily coincides with the extinction of the ances-tral species. Furthermore, because the factors affecting speciation and re-sistance to extioction in some models (i.e., the number of demes and theirspatiotemporal distributions) are not necessarily altered by speciation(e.g., Wright- 1982; Lande 1980, 1986), these models may even predictthat the ancestral lineage is more likely to speciate in the future than isthe descendant one.


-------' ,.,w V-------- W J

_______ u J

J J-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~:::::::::::::!-----j

Figure 4.6 A segment of the "triloboid" phylogeny taken from the MBL sim-ulations and stereotyping cladogenetic evolution as defined in this chapter.

As already noted, we use "bifurcation" to distinguish one of the twopatterns of speciation included within the term "cladogenesis" that is used(too broadly, in our view) by some systematists. Although Maddison(1989) referred to "multiple speciation," where one species gives rise toseveral descendants, he discussed it as an event (i.e., trifurcation, quad-rifurcation, etc.) and thus a special case of processes that produce bifur-cation (e.g., "type III" allopatry of Brooks and McLennan 1991). Nixonand Wheeler (1992) acknowledged that one daughter lineage can be in-distinguishable from the ancestral morphology, which leaves a patternindistinguishable from our definition of cladogenesis. However, they didnot discuss the possibility or likelihood of the plesiomorphic lineage pro-

. ducing additional daughter taxa.


The well-known MBL phylogeny simulation (e.g., Raup and Gould1974; Raup 1977) used cladogenesis (as we narrowly define it) exclusivelyas the pattern of speciation. Because probabilities of extinction and spe-ciation were held constant, the past history of a lineage (i.e., age ornum-ber of daughter lineages) did not affect its immediate prospects. Figure4.6 shows a segment of this phylogeny, and figure 4.7a gives the corre-sponding cladogram. Figure 4.7b-e depicts cladograms for other seg-ments of the "triloboid" phylogeny. An abundance of polytomies reflect-ing the actual relationships among "triloboid" species clearlypredominates. In addition, a strong association exists between the tem-poral range of aspecies and the number of descendants it leaves.

If modes of speciation conducive to our restricted definition of clado-genesis predominate within a clade, then an association between the ex-tinction of one species and the appearance of another is not necessarilyexpected. Based on the MBL simulation, we might expect a positive as-sociation between plesiomorphic species and either longer temporal

c

e

Figure 4.7 Cladogram accompanying figure 4.6; b-e cladograms reflect largersegments of the triloboid phylogeny than does a. Derived from Raup 1977.

96 Peter J.Wagner and Douglas H. Erwin

ranges or whatever factors are conducive to longer temporal durations.Finally, we actually expect polytomies and some "sister taxa" with dif-

ferent times of origin.

Why Cladistic Topology Is Insufficient for DiscerningPatterns of Speciation .Although cladograms are often referred to as "phylogenies," <l; true phy-l~geny is an evolutionary tree that includes a time element and ancestor-descendant hypotheses (Hennig 1966). Cladograms alone carmot discrim-inate among patterns of speciation, so discussing speciation requires morethan cladistic topologies.

Plotting (or "optimizing") character state transitions onto a cladogramis a fairly common practice. Szalay (1977) suggested that ancestral speciesshould lack apomorphies relative to their immediate descendants and thatzero-length branches imply ancestors. Some such criterion is necessary todistinguish when ancestors (and thus information about speciation pat-terns) may be present. In our own work, if a species is plesiomorphicrelative to its sister species (figure 4.8a), then we considered it as a possibleancestor. However, if both sister species are apomorphic (figure 4.8b),then we usually did not consider either species as a possible ancestor. Thesame reasoning applies to polytomies (figure 4.8c,d). A species plesio-morphic relative to a polytomous node can represent either a "hard"polytomy (Maddison 1989) or a bifurcation that includes the ancestor inthe cladogram. "Soft" polytomies, on the other hand, are artifacts ofinsufficient data and lack any phylogenetic resolution. Among fossil spe-cies, soft polytomous nodes cannot include a plesiomorphic species, andthus cannot provide useful data.

Even if a cladogram reveals plesiomorphic species, different speciationpatterns still can produce the same cladistic topologies (Paul 1985). Figure4.9 shows various anagenetic, cladogenetic, and bifurcating topologiesthat would all produce the same cladogram. The only way to distinguishthese patterns of speciation is to incorporate temporal data and convertthe cladogram into a phylogenetic tree ..

Although temporal data can disprove hypotheses predicting pseudoex-tinction (i.e., anagenesis or bifurcation), it cannot prove them. The ob-servation that a hypothesized ancestor and its descendant(s) co-occurred

a b

c d

Figure 4.8 (a) Pectinate topology with a plesiomorphic outgroup; (b) pecti-nate topology with an apomorphic outgroup; (c) poly tomy with a plesiom-orphic species; (d) polytomy with only apomorphic species. Bars representapomorphies. Cladograms a and c are potentially informative here' h ~~,..1 dare not.

VS.ml

a

b' I IFigure 4.9 (a) Anagenetic vs. cladogenetic speciation patterns that shouldproduce the same cladistic topQlogy; (b) bifurcating vs. cladogenetic speciationpatterns that also should produce the same cladistic topology.


is irrefutable evidence that the last appearance of the ancestral speciespost-dated the first appearance of its descendant(s) (Alroy 1992). In thisinstance, cladogenesis would be the only possible mode of speciation.However, the observation that a hypothesized ancestor disappears beforethe first known appearance(s) of its hypothesized descendants is support-~ing, but not conclusive, evidence of pseudoextinction.

Do We Expect To Sample Ancestral Species from theFossil Record?Paleobiologists, including us, have been far more willing to hypothesizethe presence of ancestral species than have been other evolutionary biol-ogists. One of the main assumptions behind our arguments is that ances-tral species should be fairly common in the fossil record. In this sectionwe discuss whether this assumption is realistic given how factors such asgeographic and temporal ranges affect the likelihoods of both leavingdescendants and being sampled from the fossil record.

Cladogenetic speciation models as simple as the MBL simulations oras complex as the shifting balance theory predict that species with longertemporal ranges are more likely to leave descendants. In addition, if thebasic population dynamics (e.g., larval ecology, dispersal capacity) amongspecies within a clade do not change, species with wider geographic rangesshould be more likely to leave descendants. Greater temporal and geo-graphic ranges provide both more demes and more opportunities for thosedemes to diverge from the rest of the species. Closely related species usu-ally should be subject to similar taphonomic biases. If so, then the mainfactor affecting the relative probability of anyone species being sampledfrom the fossil record is the amount of sediment in which the species canbe preserved (Signor 1985; Valentine 1989) . Thus if cladogenesis predom-inates within a clade, then the expectation is that plesiomorphic speciesshould be more common than apomorphic species in the fossil record.

Modes of speciation predicting bifurcation usually make no predic-tions about the relative temporal ranges of ancestral and descendant spe-cies. However, these models do make predictions about the relative initialgeographic ranges of species. This is demonstrated in an example recentlypublished by Brooks and McLennan (1991; see Cracraft 1989 for similarexamples). They explained the cladistic relationships among species of


a 2 3 4 5

b

x x

Figure 4.10 Cladogram (a) corresponding to several vicariant events (b) thatdivided an ancestral geographic range. Adapted from Brooks and McLennan 1991

(figure 4.16, p. 112).

freshwater fish (figure 4.10a) as)a series of vicariant events, which aredepicted in figure 4.10b. The range of the ultimate ancestor is the widest,with subsequent vicariance events progressively dividing that range. Al-though Brooks and McLennan noted that ranges can and do expand overtime, the null expectation is that the oldest species will have the widestgeographic ranges,and ancestors thus will have higher likelihoods of beingsampled-from the fossil record.

Even if we expect to see ancestors in -the fossil record, this does notdemonstrate that any particular pair of species are ancestor and descen-

1 2 2


dant. It has thus been claimed that the hypothesis that the ancestor hasthe widest geographic range is ultimately untestable (e.g., Engelmann andWiley 1977). However, if one accepts a most parsimonious topology, thenthere are situations in which the only statistically acceptable relationshipbetween sister taxa will be one of ancestor and descendant. The assertionthat sister taxa must share the same time of origin assumes not only aparticular pattern of evolution but also that ancestral species are not in-cluded in the analysis. In the resulting clad~grams, ancestors and descen-dants (if there are any such pairs in the species analyzed) will appear,rather, as sister taxa. Yet it is impossible to construct an evolutionaryscenario in which ancestors do not precede their descendants. Figure 4.11shows a hypothetical case in which a fossil species appeared significantlyearlier than its extant sister taxon (see Strauss and Sadler 1989, and Mar-shall 1990 and this volume). The only sister taxon hypothesis that cannotbe rejected here is that of ancestor and descendant (Paul 1992).

Finally, it is important to note the distinction between ancestral speciesand ancestral populations. Although no hypothesis about speciationmodes predicts that ancestral species should necessarily be rarer in thefossil record than their descendants, only models of gradual anagenesispredict that ancestral populations should be sampled relatively frequently.Other models, such as peripheral isolation, predict that ancestral popu-lations (i.e., the peripheral isolates) are restricted in space and time, andare thus poor candidates for being sam?led from the fossil record (Eld-redge and Gould 1972).

a b FKAof2

95% Error Baron FKAof2

FKAof 1

Figure 4.11 Confidence limits on two species demonstrating that it is highlyimprobable that they share the same time of origin. If it is hypothesized thatthey are sister taxa, then the only probable relationship is that species 1 in (a) isthe ancestor to species 2. "FKA" = first known appearance.


Analysis of Phylogenies of Fossil TaxaWe present two examples of phylogenies of fossil taxa using real clades.The first uses two Neogene clades of foraminifera, the Globigerinidae andthe Globorotaliidae. The phylogenetic hypotheses were the work of Ken- -nett and Srinivasan (1983 ),although the evolutionary tree was taken fromStanley, Wetmore, and Kennett (1988). The second example uses theOrdovician members of a gastropod family, the Lophospiridae. A clad-istic analysis of that taxon is presented in a paper by one of us (Wagner1995).

There are two important differences between the two analyses. First,stratophenetic techniques (e.g., Gingerich 1979) were used to reconstructthe phylogenies of the forams whereas cladistic methods were used forthat of the gastropods. Second, forams have extremely complete fossilrecords, and their sampling is as close to that of computer simulations asis possible from the fossil record; early Paleozoic gastropods, on the otherhand, have a much sparser record. If the fossil record of forams is notgood enough to execute the method suggested here, then it likely cannotbe done for any taxa. Conversely, if speciation patterns within Ordoviciangastropods can be examined with phylogenies, then it certainly should bepossible to do the same within taxa with denser fossil records (e.g., post-Paleozoic molluscs or Paleozoic brachiopods).

We subjected the phylogenies to two types of analysis. First, we ex-amined whether any of the three basic speciation patterns (cladogenesis,anagenesis, or -bifurcation) appear in significantly greater proportionsthan expected. Second, we examined whether there is any associationbetween the temporal and geographic ranges of a species, its likelihoodof leaving descendants (or how many), as well as the type of speciationthat a lineage is most likely to undergo. We used simple binomial tests toexamine the distributions of speciation patterns, and Mann-Whitney andKruskal-Wallis tests to ex<}mine the possible associations between ple-siomorphic species and temporal or geographic ranges. The latter twotests are non-parametric analogs of analysis of variance (ANOVA), whichare appropriate given that normal distributions could not be assumed forany of the variables involved. SYSTAT@)for Macintosh computers (Feld-man et al. 1988) was used to perform non-parametric tests.


Cenozoic GlobigerinidsBecause most workers use cladograms to present phylogenetic hypothe-ses, we have converted the evolutionary trees presented by Stanley, Wet-more, and Kennett (1988) into cladistic topologies for this discussion (fig-ure 4.12). The resulting phylogeny reveals 40 speciation events. Five ofthese appear to be anagenetic, whereas 35 are cladogenetic. No bifurca-tion appears. A binomial test rejects a hypothesis that cladogenesis andanagenesis should occur in roughly equal proportions. In order to pro-duce probability values greater than 0.05 from the binomial test, it mustbe postulated that anagenesis accounted for only 28% of the speciationevents.To examine the possible association between likelihood of speciation

and temporal and geographic ranges, we divided the species into threeclasses: those with no likely descendants, those with one, and those withtwo or more (most non-parametric tests become increasingly less usefulwith very small counts in categories; because very few species had more, than two descendants, we lumped all species with two or more descen-dants into the same category).Figure 4.13a shows the temporal ranges of globigerinid species opti-

mized onto the cladogram. Note that the ranges of plesiomorphic speciesare given at the nodes, whereas the ranges of apomorphic species are givenat the ends of the cladogram. Table 4.2 gives the means and standarddeviations of the temporal ranges for each group. A significant associationexists between lengthier durations of species and their likelihood of leav-ing descendants (H = 7.97, P = 0.0186). However, this association ap-plied only to species that gave rise to descendants via c1adogenesis. Ifanagenetic ancestors are excluded, the association is much more signifi-cant (table 4.3). Table 4.4 shows that anagenetic ancestors tend to haveshorter temporal ranges than have other species, although this tendencyis not significant.To examine the association between speciation and geographic ranges,

we used the number of biogeographic units occupied by a species as aproxy for geographic range (according to table 2 in Stanley, Wetmore,and Kennett 1988}.Figure 4.13b displays the relationship between ple-siomorphy and geography, revealing similar patterns to those of temporalranges. There is a strong positive association between leaving descendantsand possessing wide geographic ranges (Kruskal-Wallis H = 8.11,


0331a31b32a32b3016282418a18b27a27b0610260725212242a0842b0919

Figure 4.12 Cladogram for Neogene globigerinid foraminifera (derived fromStanley, Wetmore, and Kennett 1988). Species are numbered as in Stanley, Wet-more, and Kennett 1988, with letters denoting different morphospecies within

, '

anagenetic lineages.

43120529023304

~g

. 36141541172038373534?

mm dado genic an!x!s!()fs

o possible anngenicancestors

p = 0.0173), especially via cladogenesis (Kruskal-Wallis H = 9.76,p = 0..0076). However, there was no association between the number ofcladogenetic descendants and either temporal or geographic ranges ofancestral species (Mann-Whitney test p = 0.643).

We conducted a similar analysis for another Neogene foram clade, the

~ bol.hcladogcnicand3nagenicdescendants

~: ~~• •~.. ..~.• •• /~ .

" ....... ............ ....~ dadogenic ancestor ~~: ~ 1 ~'" '.[J .---A •• 2o anagenic ancestor ,~ r;:] 3..

'-.'-.1 •• 4[J~. .... ....@ species with anagenic and

EJ 5:-:-:

~;.....cladogenic descendants ...

~; [ill.... 6.......• apomorphic species _.:.:ss.:••

ba

•

••

Figure 4.13 Cladogram of globigerinid species (same as 4.12, but with addi-tional information depicting (a) the relative temporal ranges and (b) the relativenumber of biogeographic units occupied by each species). Geographic rangesare given at the nodes. Black circles or squares at the ends of the cladogramssignify species with no known descendants (apomorphic species). The diametersand widths of these circles and squares are proportional to the temporal andgeographic range.

•


Table 4.2Association Between Number of Descendants and Temporal Range

Among Globigerinid Foraminifera

Number Descendants Number SpeciesTemporal Range(average My)

o12+

18226

5.39 ± 4.2110.25 ± 6.9918.02 ± 11.00

SOURCE: Temporal ranges taken from Stanley, Wetmore, and Kennett 1988.NOTE: Results of Kruskal-Wallis test are H = 7.97, P = 0.0186.

Table 4.3Association Between Number of Cladogenetic Descendants

and Temporal Range among Globigerinids

Temporal RangeNumber Descendants Number Species (average My)

0 22 4.85 ± 4.091 17 12.60 ± 6.082+ 6 15.30 ± 11.88

NOTE: Association between cladogenesis and geographic range is H 9.76, P = 0.0076; results ofKruskal-Wallis test are H = 14.53, P = 0.0007.

Table 4.4.Association Between Number of Anagenetic Descendants and

Geographic Range among Globigerinids

Number Descendantsvia Anagenesis Number Species

Temporal Range(average My)

o1

405

9.73 ± 7.594.70 ± 5.61

NOTE: Association between geographic range and anagenesis is Z = -0.46, P = 0.6425; results of Mann-Whitney test are U = 88; p= 0.0577.

globorotaliids. To conserve space, the results are not presented here, butthey were si~ilar to those found for globigerinids: clad ogene sis is signif-

- icantly more common than anagenesis, a positive association exists be-tween having long temporal ranges and leaving cladogenetic descendants,and no such association exists for anagenetic ancestors. However, the

106 Peter ]. Wagner and Douglas H. Erwin

association between ancestral species and geographic ranges was not sig-nificant, even for cladogenetic ancestors.

There are no previous hypotheses about predominant modes of spe-ciation in globigerinids against which to compare our results. However,such hypotheses should predict the following: (1) a heavy predominanceof cladogenesis rather than anagenesis; (2) little or no bifurcation; (3) anincreased likelihood of cladogenetic speciation if factors encourage longertemporal ranges (or realized longer temporal ~anges); (4) a similar asso-ciation between cladogenetic speciation and wider geographic ranges; and(5) anagenetic "morphospecies" possessing shorter durations than speciesprone to cladogenesis. Hypotheses for the globorotaliids should be simi-lar, save for point 4, as there is no association between geographic rangeand leaving descendants.

Ordovician LophospiridsUnlike the foram data, all data for the Lophospiridae are inferred from acladogram (figure 4.14). As detailed elsewhere (Wagner 1995), this is notthe most parsimonious overall cladogram, but it is the most parsimoniousone that cannot be rejected by stratigraphic data. There have been noprevious phylogenetic analyses for this family, although workers havepostulated multiple descendants from a few wide-ranging species (e.g.,

, Ulrich and Scofield 1897; Tofel and Bretsky 1987).Our preferred cladogram for lophospirids (figure 4.14) is rife with po-

lytomies. Of the eleven polytomies, only two do not include plesiom-orphic species. Thus, nine may represent hard polytomies. Similarly, ofthe nine pectinate topologies, eight include plesiomorphic species. All 23apomorphic species postdate apomorphic sister lineages in the fossil rec-ord, although this is statistically significant for only 11 cases. However,in those cases, the only tenable sister-taxon hypothesis is that of ancestorand descendant.

By incorporating data as t-o stratigraphic ranges, we found that onlysix of 42 implied speciation events could have been anagenetic and thatonly one could represent bifurcation. A binomial test strongly rejects theidea that anagenesis and cladogenesis were equally likely among lopho-

- spirids (p = 1.419 X 10-6). Itmust be postulated that anagenesis accountsfor only 27% of speciations to produce probability values greater than0.05.


no evidencefor"ancestral type

P "Lophospira" perangulataA "Lophospira" kangyaoensisA "Lophospira" ventricosaA "Lophospira" serrulataP N.GenusA decursaA N.GenusA conoideaA N.GenusAproductaA N.GenusA sp.A N.GenusB spironemaP N.GenusB burgenensisP N.GenusB oweniP N.GenusB sumnerensisA N.GenusB tropidophoraA N.GenusB mooreiP Lophospira milleriA "Trochonemella" knoxvillensisP Trochonema umbilicatumA Trochonema historicumA Trochonema eccentricumA Trochonema subcrasumA Trochonema madisonenseP Trochonema bellulumP Trochonema trochonemoidesP Trochonemella montrealensisP Trochonemella sp. IA Trochonemella reusingiA Trochonemella churkiniA Trochonemella sp. IIP Trochonemella notablisP Arjamannia thraivensisA Lophospira bellicarinataP "Ruedemannia" humilisA Ruedemannia lirataA Schizolopha textilisP Eunema helicteresA Eunema strigillatumA Eunema quadrisulcataP Gyronema pulchellumA Gyronema semicarinatumP Eunema centra lisP "Lophospira" grabauiA Donaldiella bowdeniA Donaldiella filosaP Donaldiella derwiduiiP Donaldiella ?cicelia

IIoA

cladogenic ancestors

possible anagenicancestors

Figure 4.14 Cladogram for Ordovician Lophospiridae, including informationon probable patterns of cladogen~sis and possible patterns of anagenesis. P de-notes ancestral, or plesiomorphic, species; A denotes apomorphic species.

We can no~ explore the possible association between likely ancestorsand stratigraphic and geographic ranges. As with the forams, we dividedlophospirid species into three groups for"the Kruskal- Wallis test: thosewith no likely descendants, those with one possible descendant, and thosewith two or more descendants. Given the apparent predominance of c1ad-


ogenesis, the same tests for temporal or geographic association were per-formed. Figure 4.15 presents the cladogram of the lophospirids with rel-ative temporal ranges optimized onto it in the same manner as for theforam cladogram. Temporal ranges are derived from Harland et al. 1990;however, the same results were obtained using the time scales of Harland

••••

•~.•

/!!J!}). hypothesized cladogenic~ ancestor

o•possible anagenic

ancestor

apomorphic species ••Figure 4.15 Cladogram for Ordovician Lophospiridae (same as figure 4.14),depicting relative temporal ranges. The diameter of each circle is proportionalto the temporal range. See the appendix for actual temporal data.


5 , •l!3Q~

4 -I"0 •Q8tI)Q)

Q"0 3Q)-~-;::s-tI)0

2~~ • •• ... •~0

'""Q).08 1-1_. •;::sZ

o I I '•. , 1 T .• 1 1

o 10 20 30 40 50 60Temporal Range (106 yrs.)

Figure 4.16 The number of putative descendants plotted against temporal _,range for Ordovician lophospirid species. Temporal ranges derived from Harlandet al. 1990.

Table 4.5Association Between Temporal Ranges of PlesiomorphicLophospirid Gastropods and the Number of Lineages or

Clades to Which They Are Primitive

Number Descendants Number SpeciesTemporal Range(average My)

o12+

251110

2.98 + 3.295.41 + 4.33

21.90 + 15.31

NOTE: Temporal ranges are determined from Harland et al. 1990. Results of Kruskal-Wallis test are H= 20.63, p < 0.0001. A significant difference also exists between apomorphic species and allplesiomorphic ones (Mann-Whitney: U = 84.33, p = .0001).

et al. 1982 and Ross et al. 1982. Figure 4.16 plots the number of hy-pothesized des<;:endants.against the observed temporal range of the hy-

, pothesized ancestor. Table 4.5 compares the association between the tem-poral ranges of plesiomorphic sp·eciesand the number of clades or lineagesto which they are primitive. Again, a significant association exists betweenplesiomorphic species and wider temporal ranges (Kruskal-WallisH = 20.63, p < 0.0001). In general, plesiomorphic species have longer


temporal ranges than apomorphic ones (Mann-Whitney test p < 0.0001).However, among plesiomorphic species, possibly anagenetic ancestorshave significantly shorter temporal durations (Mann-Whitney testp = 0.01; table 4.6) and are not distinguishable from apomorphic species.

Figure 4.17 optimizes geographic ranges mito the cladogram. Thesewere approximated from the known distributions of lophospirid horizonsand the paleogeographic reconstructions of Scotese (1989). Unlike Neo-gene forams, the association between plesiomorphic species. and wider.geographic ranges is significant (tables 4.7 and 4.8). Among plesio-morphic species, a significant association also exists bet~een possiblecases of anagenesis and restricted geographic ranges.

The relationship between plesiomorphy and 'both temporal and geo-graphic ranges in the lophospirids we analyzed suggests that: (1) longertemporal ranges promoted both wider geographic ranges and increasedlikelihoods of cladogenesis; (2) wider geographic ranges promoted longertemporal ranges and increased the likelihood of cladogenesis; or (3) somefactor (or set of factors) encouraging longer durations and greater geo-graphic ranges of species also encouraged cladogenesis. Nearly oppositestatements can be made about the relationships between anagenesis andboth temporal and geographic ranges. Therefore, any hypothesis aboutdominant modes of speciation among Ordovician lophospirids should beable to predict the following: (1) cladogenesis is significantly more com-mon than anagenesis; (2) species with longer temporal and geographicranges are more likely to leave descendants via cladogenesis or the factorscontributing to wider temporal and geographic ranges also contribute tothe likelihood of cladogenetic evolution; (3) if anagenesis occurs, it onlyapplies to species with restricted temporal and geographic ranges; and (4)bifurcation accounts for a negligible amount of speciation.

Table 4.6Association Between Temporal Ranges of Lophospirid Gastropods and

Hypothesized Cladogenetic and Anagenetic Ancestors

Ancestral Type Number SpeciesTemporal Range(average My)

CladogeneticAnagenetic

156

17.23 ± 15.564.33 ± 2.86

NOTE: Results of Mann-Whitney test are U = 75, P = 0.01.

No. ofBiogeographic Units

o 1

o 2

0304OS AY/// ;;, .

---- - --- •••••••••••••

~ hypothesized cladogenic ,~:ancestor

0 possible anagenicancestor

II apomorphic species ~:Figure 4:17 Lophospirid cladogram depicting relative geographic ranges,as inferred from Scotese 1989. Key for geographic ranges is 1: < 104 km2;2: 104 km2 to 105 km2; 3: 105 km2 to 106 km2; 4: 106 km2 to 107 km2;5: > 107 km2. See the appendix.


Table 4.7Association Between Plesiomorphic Species and

Geographic Ranges of Lophospirids

Number Descendants Number Species

NumberGeographic Units

(average)

o12+

251110

11-23-4

NOTE: Results of Kruskal-Wallis test are H = 10.60, p = 0.005. A significant difference does not,however, exist between apomorphic species and all plesiomorphic ones (Mann-Whitney test p = 0.713).

Table 4.8Association Between Geographic Ranges and Hypothesized Cladogenetic

and Anagenetic Ancestors of Lophospirids

Ancestral Type Number Species

NumberGeographic Units

(average)

CladogeneticAnagenetic

15'6

3-41

NOTE: Results of Mann-Whitney test are p = 0.0195.

Finally, it bears noting that if the phylogeny of a taxon such as thelophospirids can limit speciation hypotheses, then the methods outlinedhere should be applicable to a wide range of fossil taxa.

Comparisons with Previous WorkOther workers have suggested associations between temporal/geographicranges and patterns of speciation (e.g., Scheltema 1978; Jablonski andLutz 1983; Jablonski 1986). For example, Scheltema (1978) predictedthat species of prosobranch gastropods ~ith planktotrophic larvae gen-erally have greater temporal and geographic ranges, and are prone tocladogenesis. Scheltema also predicted that species with nonplankto-trophic larvae are restricted in time and space, and are more prone toanagenesis; The data presented here corroborate part of this hypothesis:the association between patterns of speciation and temporal/geographicranges. However, they cannot address the association between differences


in fundamental population dynamics and speciation patterns, as there isno evidence for variation in basic population dynamics within the taxaanalyzed here.

Other workers have focused primarily on rates of speciation. For ex-ample, Jablonski (1986) found that species with nonplanktotrophic larvaewere more likely to undergo anagenesis and were prone to have fasterspeciation rates. We have not addressed the question of speciation rates,but the model presented here could be modified to examine differentialrates of evolution.

Implications for Phylogenetic Systematicsand Comparative BiologyBecause bifurcation has been assumed to be the dominant pattern of spe-ciation, many polytomies have been dismissed almost despairingly as un-resolved. However, in both examples considered here, little (if any) bi-furcation was uncovered. Imposing a bifurcating model on the data wouldresult in gross misinterpretations of the clades' evolutionary histories. Akey point of this paper is that patterns of speciation should be tested byphylogenies-not simply assumed when generating and deciphering cla-dograms. Thus, many polytomies actually may reflect evolutionary pat-terns. This represents a testable possibility that should be explored in allcases.

A similar logical problem affects recent methodological advances incomparative biology. The main point of Felsenstein 1985, that speciescannot be treated as independent entities, is well taken. However, recentphylogenetic models of comparative biology (e.g., Maddison 1990; Har-vey and Pagel 1991; Martins and Garland 1991) generally assume thatevolution during speciation is no different from evolution within species.Whenever possible this assumption should be tested by phylogenies ratherthan assumed when those phylogenies are used in comparative biology.Such testing probably is not possible for studies using taxa with poorfossil records or concerning traits that do not fossilize (e.g., Coddington1988; Gittleman 1986a, 1986b; Losos 1990, 1992; Lynch 1991). How-ever, given a phylogenetic hypothesis, comparative biological studies us-ing morphologic data can test whether intraspecific evolution is the sameas interspecific evolution. The one such study to date disputes this as-sumption (Cheetham 1986, 1987; Cheetham and Jackson, this volume;


but also see Marshall, this volume). More such studies are needed beforegeneralizations can be made, but in all cases paleobiologists should testand justify the assumptions of the particular comparative method thatthey use.

Applicability and LimitationsOnly species-level phylogenies should be analyzed in.the manner set forthin this paper. Genus or family level phylogenies omit too many speciationevents and cannot exactly describe the timing 6f the included events.Moreover, it is possible that morphologic changes great enough to berecognized as new genera or families do not have the same type of dis-tribution within a phylogeny as do standard speciation events. We arenot claiming that this possible distinction between specific and supra-specific evolution is necessarily true, but it is something that should betested using species-level phylogenies. Additionally, clades should be wellsampled at the species level. Good sampling not only increases the ro-bustness of a phylogenetic hypothesis (Paul 1982, 1985; Allmon 1989)but also diminishes the tricky problems of gaps that accompany analysesof evolutionary relationships above the species level.

In the introduction, we mentioned that phylogenies can corroborateor falsify hypotheses about speciation modes. However, the methods dis-cussed here cannot demonstrate whether a particular hypothesis is cor-rect. More than one mode of speciation may be consistent with a partic-ular phylogenetic result, or the fossil record may not allow discriminationbetween competing hypotheses. The modes of speciation producing clad-ogenesis, anagenesis, and perhaps even bifurcation can be either punc-tuated or gradual (MacLeod 1991). Well-documented phylogenies canshow stasis within lineages (in a loose sense, not sensu Bookstein 1987,1988) if MBL-type patterns of evolution are demonstrated. However, aphylogeny cannot demonstrate whether the cladogenetic speciation wasgradual or punctuated.

Although phylogenies and cladograms cannot reveal modes of speci-ation, there is a danger that they create the impression of o~ly punctuatedchange (Gayon 1990). Traditional ("alpha") systematics relies on a com-bination of both discrete, meristic characters (i.e., things one can count)and continuously variable characters (i.e., shapes and sizes). In contrast,cladistic analyses require that both types of characters be described as ifthey were discrete. This is not because systematists are ignorant of the


importance of continuously variable characters, but simply because reli-able and repeatable coding of differences in shape and size is a very dif-ficult problem (e.g., see Thorpe 1984; Archie 1985; Goldman 1988).

The potential bias toward perceiving punctuated change may be ex-acerbated by the fact that a cladogram contains no information regardingthe timing of character state changes. A hypothetical example is given infigure 4.18. If specimens are sampled and coded from three points alonga gradually evolving lineage (figure 4.18a), the correct cladogram (figure

f- sample#3

I1,) ?t"\-":-Of)

;. '-~ ."

turnover ~s::c:!..s::u f- sample#2V)::l0::l

'~I• turnover ~0u f- sample#l

a b

Figure 4.18 An example of how gradual and punctuated evolution (sensuMacLeod 1991) cannot be distinguished on a cladogram. A cladistic analysIs ofspecimens from the three points along a gradually anagenetic lineage (a) shouldlead to the cladogram shown in c. This cladogram could inadvertently give theimpression that punctuated anagenetic change occurred (b), despite lack of sup-porting evidence.


4.18c) will link the later specimens as sister taxa using qualitative statesof continuously evolving characters. A literal reading of this may give theimpression that punctuated anagenesis occurred (figure 4.18b), but thiswould be purely an artifact of the methods of phylogenetic analysis.

Hypotheses about modes of speciation make predictions about" pat-terns of speciation, which in turn make predictions about patterns withinphylogenies. Therefore, phylogenetic patterns off~r a potential test of spe-ciation hypotheses that can corroborate or disprove (but, alas, not prove)those hypotheses. Two examples we have drawn from the fossil recordindicate that testing of speciation hypotheses can be done not only fortaxa with excellent fossil records but also for those with more averagerecords. Thus, even if the fossil record of a clade leaves no direct evidenceabout common modes of speciation, the clade's phylogenetic patterns canbe used to limit the appropriate hypotheses.

Appendix

Temporal and Geographic Ranges of Lophospirid Species and the Numberof Lineages to Which They Were Plesiomorphic

Temporal ranges are derived from Harland et al. 1990. Geographic rangeswere approximated using the reconstructions in Scotese 1989. See the captionof figure 4.17 for the key to the geographic ranges signified by the numbersthat appear in that figure and here. A. = Arjamannia; D. = Donaldiella; E.= Eunema; C. = Gyronema; L. = Lophospira; R. = Ruedemannia; S.Schizolopha; Ta. = Trochonema; Tl. = Trochonemella.

Species Temporal Geographic PossibleRange (My) Range Descendants

D. ?cicelia 29 4 2"L." perangulata 50 5 5"L." kangyaoensis 12 2 0"L." grabaui 16 3 1L. milleri 37 5 4D. derwiduii 8 1 1"Tl." knoxvillensis 13 1 0E. centralis 21 2 2Ta. trochonemoides 1.5 1 1Ta. bellulum 6 1 2Ta. eccentricum 4 1 0


Tl. montrealensisE. helicteres"L." ventricosa"L." serrulataTI. sp. ITa. umbilicatumE. strigillatumTl. notabilisG. pulchellumG. semi carina tumN. Gen. B burgenensisN. Gen. B spironemaN. Gen. B oweniN. Gen. B sumnerensisTa. historicumTa. subcrasum"R." humilisN. Gen. Asp.N. Gen. A decursaN. Gen. A conoideaN. Gen. A productaS. textilisD. bowdeniD. filosaN. Gen. B tropidophoraR. lirataL. aff. L. serrulataTI. churkiniTI. reusingiL. bellicarinataE. quadrisulcataA. thraivensisN. Gen. B mooreiTa. madisonense

313

552.51952.582

231102386121116.515311133412

1111121111422111211111112111111311

12oo12o11o2o2oo12o1ooooo1oooo1oooo

Acknowledgments P. W. would like to thank the editors of this volume firstfor the invitation to present a paper at the 1992 G.S.A. symposium they organ-ized, and second for the invi~ation to be a contributor (with coauthor DouglasErwin). This paper originated from a class project by P. W. concerning biases inthe fossil record, which was undertaken for David Raup's "Evolutionary Paleo-biology" course. Thanks are owed to him for making the original MBL notesavailable. Both authors would like to thank the following people who critiqued.various portions and iterations of the manus·cript or provided valuable discus-sion: J. Alroy, G. Eble, D. Jablonski, E. E. LeClair, R. Lupia, D. M. Raup, andJ. J. Sepkoski, Jr.


References

Allmon, W. D. 1989.Paleontological completeness of the record of lower TertiaryMollusks, U.S. Gulf and Atlantic coastal plains: Implication for phylogeneticstudies. Historical Biology 3:141-158.

Alroy, ]. 1992. Conjunction among taxonomic distributions and. the Miocenemammalian biochronology of the Great Plains. Paleobiology 18:326-343.

Archie,]. W. 1985.Methods for coding variable morphological features for nu-merical taxonomic analysis. Systematic Zoology 34:326-345.

Bookstein, F. L. 1987. Random walk and the existence of evolutionary rates.Paleobiology 13:446-464. .

---. 1988. Random walk and the biometrics of morphologic characters. Ev-olutionary Biology 23:369-398.

Boucot, A.]. 1978. Community 'evolution and rates of cladogenesis. EvolutionaryBiology 11~545-655.

Brooks, D. R. and D. A. McLennan. 1991. Phylogeny, ecology and behavior: Aresearch program in comparative biology. Chicago: University of ChicagoPress.

---.1986. Tempo of evolution in a Neogene bryozoan: Rates of morphologicchange within and across species boundaries. Paleobiology 12:190-202.

Cheetham, A. H. 1987. Tempo of evolution in a Neogene bryozoan: Are trendsin single morphologic characters misleading? Paleobiology 13:286-296.

Coddington,]. A. 1988.Cladistic tests of adaptational hypotheses. Cladistics 4:3-22.

Cracraft,]. 1981. Pattern and process in paleobiology: The role of cladistics insystematic paleontology. Paleobiology 7:456-468.

---. 1989. Speciation and its ontology: The empirical consequences of alter-native species concepts for understanding patterns and processes of differen-tiation. In'D. Otte and]. A. Endler, eds., Speciation and Its Consequences, pp.28-59. Sunderland, Mass.: Sinauer.

Darwin, C. 1859. The Origin of Species. New York: Random House.Eldredge, N. 1971. The allopatric model and phylogeny in Paleozoic inverte-

brates. Evolution 25:156-167.Eldredge, N. and S.]. Gould. 1972. Punctuated equilibria: An alternative to phy-

letic gradualism. In T.]. M. Schopf, ed., Models in Paleobiology, pp. 82-115.San Francisco: Freeman.

Engelmann, G. F. and E. O. Wiley. 1977. The place of ancestor-descendant re-lationships in phylogeny reconstruction. Systematic Zoology 26:1-11.

Feldman, D., ]. Gagne, R. Hofmann, and]. Simpson. 1988. StatvieuJiIfJ SE +Graphics 1.02. Berkeley: Abacus Concepts.

Felsenstein,]. 1985. Phylogenies and the comparative method. American Natu-ralist 125:1-15.

Gayon,]. 1990. Critics and criticisms of the modern synthesis-the viewpoint ofa philosopher. Evolutionary Biology 24:1-49.


Gingerich, P. D. 1976. Paleontology and phylogeny: patterns of evolution at thespecies level in early Tertiary mammals. American Journal of Science 276:1-28.

---. 1979. The stratophenetic approach to phylogeny reconstruction in ver-tebrate paleontology. In J. Cracraft and N. Eldredge, eds., Phylogenetic Anal-ysis and Paleontology. pp. 41-77. New York: Columbia University Press.

--~. 1985. Species in the fossil record: Concepts, trends, and transitions. Pa-leobiology 11:27-42.

Gittleman, J. L. 1986a. Carnivore brain size, behavioral ecology, and phylogeny.Journal of Mammalogy 67:23-36.

---. 1986b. Carnivore life history patterns: Allometric, phylogenetic, and eco-logical associations. American Naturalist 127: 744-771.

Goldman, N. 1988. Methods for discrete coding of morphological characters fornumerical analysis. Cladistics 4:59-71.

Guyer, C. and J. B. Slowinski. 1991. Comparisons of observed phylogenetic to-pologies with null expectations among three monophyletic lineages. Evolution45:340-350.

Harland, W. B., R. L. Armstrong, A. V. Cox, L. E. Craig, A. G. Smith, and D.G. Smith 1990. A Geologic Time Scale 1989. Cambridge: Cambridge Univer-sity Press.

Harland, W. B., A. V. Cox, P. G. Llewellyn, C. A. G. Pickton, A. G. Smith, andR. Walters. 1982. A Geologic Time Scale. Cambridge: Cambridge UniversityPress.

Harvey, P. H. and M. D. Pagel 1991. The Comparative Method in EvolutionaryBiology. Oxford: Oxford Press.

Heard, S. B. 1992. Patterns in tree balance among cladistic, phenetic, and ran-domly generated phylogenetic trees. Evolution 46:1818-1826.

Hennig, W. 1966. Phylogenetic Systematics. Urbana: University of Illinois Press.Jablonski, D. 1986. Larval ecology and macroevolution in marine invertebrates.

Bulletin of Marine Science 39:565-587.Jablonski, D. and R. A. Lutz. 1983. Larval ecology of marine benthic inverte-

brates: Paleobiological implications. Biological Review 58:21-89.Kennett, J. P. and M. S. Srinivasan. 1983. Neogene Planktonic Foraminifera-a

Phylogenetic Atlas. Stroudsburg, Penn.: Hutchinson Ross.Lande, R. 1980. Genetic variation and phenotypic evolution during peripheral

isolation. American Naturalist 116:463-479.---.1982. Rapid origin of sexual isolation and character divergence in a cline.

Evolution 36:213-223.---. 1986. The dynamics of peak shifts and the pattern of morphologic evo-

lution. Paleobiology 12:343-354.Larson, A. 19..89.The relationship between speciation and morphologic evolution.

In D. Otte and J. A. Endler, eds., Speciation and Its Consequences, pp. 579-598. Sunderland, Mass.: Sinauer.

Lorenzen, S. and J. Sieg. 1991. PHYLIP, PAUP, and HENNIG86-how reliable


are computer parsimony programs used in systematics? Zeitschrift fur Zool-ogisches Systematische Evolution 22:234-263.

Losos,]. B. 1990.A phylogenetic analysis of character displacement in CaribbeanAnolis lizards. Evolution 44:558-569.

---. 1992. The evolution of convergent structure in Caribbean Anolis com-munities. Systematic Biology 41:403-420.

Lynch, ]. D. 1989. The gauge of speciation: On the frequenci~s of modes ofspeciation. In D. Otte and J. A. Endler, eds., Speciation and Its Consequences,pp. 527-553. Sunderland, Mass.: Sinauer.

Lynch, M. 1991. Methods for the analysis of comparative data in evolutionarybiology. Evolution 45:1065-1080. ,

MacLeod, N. 1991. Punctuated anagenesis and the importance of stratigraphy topaleobiology. Paleobiology 17:167-188.

Maddison, W. P. 1989. Reconstructing character evolution on polytomous cla-dograms. Cladistics 5:365-377.

---. 1990. A method for testing the correlated evolution of two binary char-acters: Are gains or losses concentrated on certain branches of a phylogenetictree? Evolution 44:539-557.

Maddison, W. P. and M. Slatkin. 1991. Null models for the number of evolu-tionary steps in a character on a phylogenetic tree. Evolution 45:1184-1197.

Marshall, C. R. 1990. Confidence intervals on stratigraphic ranges. Paleobiology16:1-10.

Martins, E. P. and T. Garland ]r. 1991. Phylogenetic analysis of the correlatedevolution of continuous characters: A simulation study. Evolution 45:534-557.

Maynard Smith, ]. 1966. Sympatric Speciation. American Naturalist 100:637-650.

Mayr, E. 1963. Animal Species and Evolution. Cambridge: Harvard UniversityPress.

---. 1982. Processes of speciation in animals. In C. Barigozzi, ed., Mechanismsof Speciation, pp. 1-19. New York: Alan R. Liss Press.

Nixon, K. C. and Q. D. Wheeler. 1992. Extinction and the origin of species. InM. ]. Novacek and Q. D. Wheeler, eds., Extinction and Phylogeny, pp. 119-143. New York: Columbia University Press.

Panchen, A. L. 1992. Classification, Evolution and the Nature of Biology. Cam-bridge, U.K.: Cambridge University Press.

Paul, C. R. C. 1982. The adequacy of the fossil record. In K. A. ]oysey and A. E.Friday, eds., Problems of Phylogenetic Reconstruction, pp. 75-117. London:Academic Press.

---. 1985. The adequacy of the fossil record revisited. Special Papers in Pa-laeontology 33:1-16.

---. 1992. The recognition of ancestors. Historical Biology 6:239-250.Provine, W. B. 1989. Founder effects and genetic revolutions in microevolution

and speciation: A historical perspective. In L. V. Giddings, K. Y. Ashiro, and


W. W. A. Anderson, eds., Genetics and the Founder Principle, pp. 43-60.London: Oxford University Press.

Raup, D. M. 1977. Stochastic models in evolutionary palaeontology. In A. Hal-lam, ed., Patterns of Evolution, pp. 59-78. Amsterdam: Elsevier.

Raup, D. M. and S. J. Gould 1974. Stochastic simulation and evolution of mor-phology-towards a nomothetic paleontology. Systematic Zoology 23:305-322.

Rensch, B. 1959. Evolution Above the Species Level. New York: Columbia Uni-versity Press.

Ross, R. ]. ]., F. ]. Adler, T. W. Amsden, D. Bergstrom, S. M. Bergstrom, C.Carter, M. Churkin, E. A. Cressman,]. R. Derby,]. T. Dutro, R. L. Ethington,S. C. Finney, D. W. Fisher, ]. H. Fisher, A. G. Harris, L. F. Hintze, K. B.Ketner, K. L. Kolata, E. Landing, R. B. Neuman, W. C. Sweet, ]. Pojeta, ]r.,A. W. Potter, E. K. Rader,]. E. Repetski, R. H. Shaver, T. L. Thompson, andG. F. Webers. 1982. The Ordovician System in the United States. InternationalUnion of Geological Sciences.

Scheltema, R. S. 1978. On the relationship between dispersal of pelagic veligerlarvae and the evolution of marine prosobranch gastropods. In B. Battagliaand]. A. Beardmore, eds., Marine Organisms: Genetics, Ecology and Evolu-tion, pp. 303-322. New York: Plenum Press.

Scotese, C. R. 1989. Phanerozoic reconstructions: A new look at the assembly ofAsia. Phanerozoic Mapping Project Progress Report 19:1-10.

Signor, P. W. 1985. Real and apparent trends in species richness through time.In]. W. Valentine, ed., Phanerozoic Diversity Patterns: Profiles in Macroev-olution, pp. 129-150. Princeton: Princeton University Press.

Slowinski, ]. B. and C. Guyer 1989a. Testing null models in questions of evolu-tionary success. Systematic Zoology 38:189-191.

--~ .. 1989b. Testing the stochasticity of patterns of organismal diversity: Animproved null model. American Naturalist 134:907-921.

Smith, G. R. 1992. Introgression in fishes: Significance for paleontology, cladis-tics, and evolutionary rates. Systematic Biology 41:41-57.

Stanley, S. M., K. L. Wetmore, and]. P. Kennett. 1988. Macroevolutionary dif-ferences between two major clades of Neogene planktonic foraminifera. Pa-leobiology 14:235-249.

Strauss, D. and P. M. Sadler. 1989. Classical confidence intervals and Bayesianprobability estim~tes for ends of local taxon ranges. Mathematical Geology21:411-427.

Szalay, F. S.1977. Ancestors, descendants, sister groups, and testing of phylo-genetic hypotheses. Systematic Zoology 26:12-18.

Thorpe, R. S. 1984. Coding morphometric characters for constructing distanceWagner networks. Evolution 38:244-255.

Tofel, ]. E. and P. W. Bretsky. 1987. Middle Ordovician Lophospira (Archaeo-gastropoda) from the upper Mississippi Valley. Journal of Paleontology61:700-723.


Ulrich, E. O. and W. H. Scofield. 1897. The Lower Silurian Gastropoda of Min-nesota. Paleontology of Minnesota 3:813-1081.

Valentine, J. W. 1989. How good was the fossil record? Clues from the CaliforniaPleistocene. Paleobiology 15:83-94.

Wagner, P. J. 1995. Stratigraphic tests of cladistic hypotheses, Paleobiology 21.Wake, D. B., K. P. Yanev, and M. M. Frelow. 1989. Sympatry and hybridization

in a "ring species": The plethodontid salamander En$atina eschscholtzii. In D.·Otte and ]. A. Endler, eds., Speciation and Its Consequences, pp. 134-157.Sunderland, Mass.: Sinauer.

Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97-159.-.,---,. 1932. The roles of mutation, inbreeding, crossbreeding, and selection in

evolution. Proceedings of the Sixth International Congress of Genetics 1:356-366.

---. 1982. Character change, speciation, and the higher taxa. Evolution36:427-443.

Phylogenetic tests of speciation hypotheses

Documents

Transcript of Phylogenetic tests of speciation hypotheses