© 2006 Society of Systematic Biologists
Ancestral State Reconstruction of Body Size in the Caniformia (Carnivora, Mammalia): The Effects of Incorporating Data from the Fossil Record
Edited by Todd Oakley: Associate Editor
1 Committee on Evolutionary Biology, The University of Chicago 1025 E. 57th Street, Chicago, Illinois 60637, USA E-mail: johnf{at}uchicago.edu
2 Department of Geology, The Field Museum 1400 S. Lake Shore Drive, Chicago, Illinois 60605, USA
3 Division of Paleontology, American Museum of Natural History Central Park West at 79th Street, New York, New York 10024, USA
 |
Abstract |
|---|
 Top Abstract BackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
A recent molecular phylogeny of the mammalian order Carnivora implied large body size as the ancestral condition for the caniform subclade Arctoidea using the distribution of species mean body sizes among living taxa. "Extant taxa–only" approaches such as these discount character state observations for fossil members of living clades and completely ignore data from extinct lineages. To more rigorously reconstruct body sizes of ancestral forms within the Caniformia, body size and first appearance data were collected for 149 extant and 367 extinct taxa. Body sizes were reconstructed for four ancestral nodes using weighted squared-change parsimony on log-transformed body mass data. Reconstructions based on extant taxa alone favored large body sizes (on the order of 10 to 50 kg) for the last common ancestors of both the Caniformia and Arctoidea. In contrast, reconstructions incorporating fossil data support small body sizes (< 5 kg) for the ancestors of those clades. When the temporal information associated with fossil data was discarded, body size reconstructions became ambiguous, demonstrating that incorporating both character state and temporal information from fossil taxa unambiguously supports a small ancestral body size, thereby falsifying hypotheses derived from extant taxa alone. Body size reconstructions for Caniformia, Arctoidea, and Musteloidea were not sensitive to potential errors introduced by uncertainty in the position of extinct lineages relative to the molecular topology, or to missing body size data for extinct members of an entire major clade (the aquatic Pinnipedia). Incorporating character state observations and temporal information from the fossil record into hypothesis testing has a significant impact on the ability to reconstruct ancestral characters and constrains the range of potential hypotheses of character evolution. Fossil data here provide the evidence to reliably document trends of both increasing and decreasing body size in several caniform clades. More generally, including fossils in such analyses incorporates evidence of directional trends, thereby yielding more reliable ancestral character state reconstructions.
Keywords: Ancestral state reconstruction; Arctoidea; body size; Caniformia; Carnivora; evolutionary trends; Musteloidea; weighted squared-change parsimony
Received March 4, 2005; Revised August 8, 2005; Accepted November 15, 2005
Ancestral character state reconstruction can aid in understanding both the pattern and timing of character evolution (Cunningham et al., 1998; Maddison, 1995; Schluter et al., 1997; Swofford and Maddison, 1987). Such analyses often make use of well-supported phylogenies of extant taxa without reference to potentially relevant character information documented in the fossil record (e.g., Gittleman and Purvis, 1998; Harvey et al., 1994; Pagel et al., 2004; Webster et al., 2004; Webster and Purvis, 2002a). However, it has been noted that reconstructions of ancestral character states tend to become less reliable toward the root of the tree, even when the phylogeny is completely known (Oakley and Cunningham, 2000), although increased taxon sampling has been shown to increase the accuracy of root and internal node estimation (Maddison, 1995; Martins, 1999; Salisbury and Kim, 2001).
Several empirical studies have examined the behavior of character state reconstruction methods. In those studies, reconstructed values of ancestral nodes derived from observations of tip taxa were compared to values observed at or near ancestral nodes. Oakley and Cunningham (2000) used a bacteriophage system with a completely known phylogeny and a "fossil" record with known ancestral character states to demonstrate that character state reconstructions were prone to inaccuracy when directional evolutionary trends were encountered. Webster and Purvis (2002b) encountered similar inaccuracy when comparing reconstructions for various size and shape characters in planktonic Foraminifera to fossils preserved in deep sea cores. This, too, may have been the result of an observed trend toward increased size. Evolutionary trends can reduce the accuracy of character state reconstructions, especially for methods assuming Brownian motion as the model for character change. This is because an estimated root value under such a model will always be some form of weighted average of observed values for terminal taxa (Schluter et al., 1997), and if a trend moves the range of observed character state values beyond the ancestral condition, it will be difficult, if not impossible, to accurately reconstruct the condition at the ancestral node (Garland et al., 1999; Oakley and Cunningham, 2000).
As a consequence, it might seem that accurate ancestral character state reconstructions would be difficult to achieve in many real systems. It is possible, however, to incorporate information from the fossil record, for which taxa at or near ancestral positions may be most useful in reconstructions. Fossil taxa provide at least two important sources of information for ancestral reconstructions, especially with regards to previously observed weaknesses attributed to evolutionary trends. First, fossil data can more completely describe the range and distribution of character states through the entire evolutionary history of a clade. Second, the temporal information associated with character state observations in the fossil record can be used to weight the impact of individual observations on nodal reconstructions. For example, Polly (2001) used Paleogene viverravid carnivoramorphans as a model system for ancestral reconstructions of body size. By including data for presumed ancestral taxa, that study demonstrated the impact of incorporating temporal data (branch lengths) into reconstructions. By varying the length of branches to represent differential potential for evolutionary change (which could represent time, sequence divergence, etc.), character observations will "pull" reconstruction values for temporally more proximate nodes closer to the observed values (Garland et al., 1999). Thus, incorporating data from fossil taxa into character state reconstructions, especially if directional trends are observed or suspected, has the potential to greatly improve ancestral reconstructions.
The phylogeny of the living members of the mammalian order Carnivora has received considerable recent attention (Bininda-Emonds et al., 1999; Flynn and Nedbal, 1998; Flynn et al., 2000; Gaubert and Veron, 2003; Ledje and Arnason, 1996a, 1996b; Sato et al., 2003, 2004; Veron et al., 2004; Yoder et al., 2003; Yu et al., 2004a, 2004b). Recently, Flynn et al. (2005) were able to resolve the phylogenetic relationships among the major monophyletic lineages comprising the suborder Caniformia. As such, assessment of the evolution of body size in caniform carnivorans, in light of a better understanding of carnivoran phylogeny, offers an opportunity to empirically explore the effects on ancestral reconstructions resulting from the incorporation of data from the fossil record. Body size evolution is of particular biological interest, as it has been correlated with many life history and physiological variables across mammalian species in general, and carnivorans in particular (Gittleman, 1986b, 1991, 1993; Gittleman and Harvey, 1982; Kozlowski and Weiner, 1997; Schmidt-Nielsen, 1984). An enhanced understanding of body size evolution in this clade therefore has the potential to expand our knowledge of the evolution of life history, ecology, and adaptation to environment (Gittleman, 1993; Meiri et al., 2004a, 2004b), attributes that are often difficult or impossible to observe directly in the fossil record.
|
Background |
|---|
|
TopAbstractBackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
The order Carnivora includes approximately 270 known extant species, grouped into two major lineages, the cat-like Feliformia and dog-like Caniformia (Flynn and Galiano, 1982; Flynn and Wesley-Hunt, 2005; Wozencraft, 1993; Wyss and Flynn, 1993). Flynn et al. (2005) performed a comprehensive phylogenetic analysis of 76 extant carnivoran species using concatenated nucleotide data from six independent loci. That study resolved the phylogenetic relationships among the four major monophyletic lineages of the Caniformia: Canidae (dogs), Ursidae (bears), Pinnipedia (seals, sea lions and the walrus), and Musteloidea (skunks, raccoons, weasels and other allied taxa), with canids forming the sister group to all other extant caniforms, the Arctoidea. The basal divergence among arctoids splits the ursids from a clade uniting the pinnipeds and musteloids (Fig. 1).
|
Extant carnivorans span over 4 orders of magnitude of body size, and this entire range of body sizes is realized in the Caniformia. Among the major caniform lineages, extant ursids and pinnipeds are large-bodied mammals (median body sizes of
104 kg and 145 kg, respectively; Table 1), whereas musteloids are generally small (median body size 1.5 kg: Table 1). Optimizing a body size character for the living taxa across the Flynn et al. (2005) phylogeny implies a large body size as the primitive state for Arctoidea, and potentially Caniformia as well. Flynn et al. (2005) noted a significant, negative correlation between rank body size and clade rank using Spearman rank correlation, potentially implying that large body size was ancestral for caniforms and arctoids, with a subsequent body size decrease in later diverging musteloid clades.
|
Several studies have attempted to use the improved understanding of carnivoran phylogeny to reconstruct character evolution (e.g., Lindenfors et al., 2003), including body size and rates of body size evolution (Webster et al., 2004; Webster and Purvis, 2002a). However, those studies did not incorporate fossil data. Webster et al. (2004) cited concerns over incompleteness of the fossil record, which led to their "pragmatic" exclusion of fossil taxa and adoption of a "cross-sectional" (extant taxa–only) approach. Given that the aim of that study was to compare relative rates of body size evolution across clades for ancestor-descendant pairs, and that the diagnosis of true ancestor-descendant relationships must be uncertain (Smith, 1994), these concerns may be justified for that particular application.
However, extant taxa represent only a subset of the total evolutionary diversity of any clade, and, therefore, any conclusion based solely on extant taxa must be viewed with caution. Such inferences necessarily make an implicit, but problematic, assumption that character distributions among living taxa faithfully record patterns of character evolution throughout the history of the group. As a cautionary example from this study, extinct members of the Ursidae are estimated to range in body size from slightly less than 1 kg (Parictisparvus) to greater than 850 kg (Indarctos atticus). Contrasting this to the much more restricted range of species mean body masses for extant bears ( 45 to 400 kg; Table 1) emphasizes the difficulty in relying solely on recent taxa to infer patterns of body size evolution, due to the loss of information when excluding fossil taxa.
To test whether the inferred pattern of a large ancestral body size is an artifact of sampling only the extant subset or if it represents the actual evolution of body size among caniform carnivorans, ancestral body sizes for four caniform clades were reconstructed by explicitly incorporating body mass estimates for fossil taxa. Body masses were determined for fossil taxa primarily using dental proxies. Ancestral body size was reconstructed for continuous-valued body mass data using weighted squared-change parsimony. Tested in this study were (1) the importance of fossil character information for ancestral state reconstruction, by comparing "extant-only" and "all-taxa" data sets; (2) the value of temporal information for ancestral state reconstruction, through differential coding of branch lengths; and (3) the sensitivity of reconstructions to potential uncertainties in phylogeny and missing data for certain fossil lineages.
|
Methods |
|---|
|
TopAbstractBackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
Phylogeny
The interrelationships of family-level clades in the Flynn et al. (2005) phylogeny formed the backbone for the present study. That and other molecular phylogenies have revealed that many of the traditionally defined taxonomic groupings are not monophyletic. For example, skunks have historically been placed within Mustelidae and the red panda (Ailurus fulgens) often has been placed within the Procyonidae, but neither is most closely related to members of those groups. Among living taxa, a monophyletic Mephitidae and a monotypic Ailuridae form close outgroups to a clade uniting Procyonidae and a restricted Mustelidae (Dragoo and Honeycutt, 1997; Flynn et al., 2000, 2005; Ledje and Arnason, 1996a). Further, some of the traditional subfamilies within the restricted Mustelidae have been shown to be paraphyletic: for example, several taxa in the "Mustelinae" are more closely related to the Lutrinae (otters) than to other mustelines (Flynn et al., 2005; Koepfli and Wayne, 1998; 2003; Sato et al., 2003, 2004). The phylogenetic tree used here for living taxa reflects these findings (Fig. 2).
|
Fossil taxa were added to this tree, incorporating phylogenetic hypotheses for major subgroups whose monophyly and phylogenetic position are reasonably certain. These included relationships among subfamilies within Amphicyonidae (Hunt, 1996), and relationships among subfamilies and tribes within Ursidae (Hunt, 1998b) and Canidae (Tedford et al., 1995; Wang, 1994; Wang et al., 1999). Relationships among pinnipedimorph family-and subfamily-level clades (Démére et al., 2003) were also incorporated, with the exception that Odobenidae (walruses) was allied with Otariidae (sea lions), not Phocidae ("true" seals), following most recent molecular phylogenies (Davis et al., 2004; Flynn et al., 2005; Flynn and Nedbal, 1998; Ledje and Arnason, 1996a, 1996b). Within each of these broad lineages (e.g., the subfamily Borophaginae within Canidae) phylogenetic assumptions were kept to a minimum, leaving taxa in polytomies representing ambiguity in the knowledge of the true branching order (Fig. 2). The resulting analysis spans 516 caniform taxa, including 367 fossil taxa.
An incorrect phylogeny obviously can confound ancestral state reconstruction (Garland and Diaz-Uriarte, 1999; Purvis et al., 1994; Webster and Purvis, 2002b), which could present particular problems for analyses such as the present study where fossil taxa are being incorporated into a phylogenetic topology recovered by molecular data. To contend with this, the sensitivity of body size reconstructions can be evaluated relative to a suite of alternative hypotheses encompassing a range of reasonable potential positions for fossil groups whose affinities are not precisely known. For this data set, there are two such problems. First the extinct Amphicyonidae ("bear-dogs") has historically been included, either in part or wholly, within Canidae (Matthew, 1924) or Ursidae (Ginsburg, 1966; Hough, 1948). The monophyly of this clade is well established (Hunt, 1974, 1977), although its position relative to the other major caniform lineages remains controversial (Viranta, 1996). Most current phylogenies place amphicyonids as a distinct family, within or allied with the Arctoidea (Hunt, 1996, 1998a) and potentially as the sister group to ursids (Hunt, 1977; Wyss and Flynn, 1993). Recently, Wesley-Hunt and Flynn (2005) recovered support for Amphicyonidae as the sister clade to all extant caniforms. Amphicyonids therefore can be placed in several alternate phylogenetic positions: as the sister group to (1) canids, (2) ursids, (3) all extant arctoids, or (4) all extant caniforms (Fig. 3).
|
The second area of phylogenetic uncertainty concerns the "Paleomustelidae," a likely paraphyletic assemblage of early Miocene taxa, usually considered stem mustelids (Baskin, 1998). However, removing skunks from Mustelidae calls into question the phylogenetic position of these fossils. "Paleomustelids" could still be interpreted as stem taxa of a restricted Mustelidae; that is, "paleomustelids" are more closely related to taxa in the restricted Mustelidae than they are to procyonids (Fig. 3). Alternatively, "paleomustelids" could be basal to all traditional "mustelid" taxa (skunks plus the restricted Mustelidae), and therefore also basal to procyonids. In this case they would be more appropriately considered stem musteloids (Fig. 3). The position of these taxa could have a significant impact on the interpretation of body size evolution, as several "paleomustelid" taxa are large in comparison to most living musteloids and appear early in the musteloid fossil record. These two alternate hypotheses for the "Paleomustelidae" represent end-member possibilities for the phylogenetic position of these taxa.
Body Mass and Temporal Data
Body masses for fossil caniforms were estimated using regressions of area measurements (length x width) for the lower carnassial (m1), obtained from published descriptions of fossil material and from a compilation of measurement data in the North American Fossil Mammal Systematics Database (NAFMSD; Alroy, 2002). Separate regressions were constructed for the clades Canidae and Musteloidea using data from Legendre and Roth (1988), reanalyzed using the topology in Flynn et al. (2005).
Body mass estimation for extinct ursids, however, proved problematic. Fossil ursids likely spanned a greater absolute size range than any other terrestrial caniform clade. A regression for the family Ursidae derived from carnassial area produced estimates that are unrealistically large for taxa at the high end of the body mass distribution. For example, using a regression for modern Ursidae, the estimated body size of the extinct bear Indarctosatticus was in excess of 3000 kg. Although I. atticus is certainly one of the largest terrestrial carnivorans to have existed, a body mass in excess of three metric tons is highly improbable. An alternate regression based on carnassial length (Van Valkenburgh, 1990) was investigated; however, this produced estimates that were likely too large for the smaller taxa in this clade. The estimated body mass for Parictisparvus using the regression on carnassial length was 46 kg, despite the fact that its lower carnassials are smaller than those of the modern striped skunk (Mephitis mephitis; Gittleman and Van Valkenburgh, 1997), which has a body mass of only about 2 kg (Smith et al., 2003). Thus, extant ursids scale body mass to dental measures differently than do fossil members of this clade. The Amphicyonidae clade also presents a problem, as there are no living representatives to constrain estimates. Body masses were estimated for both Amphicyonidae and Ursidae using carnassial areas regressed across all caniform taxa (Viranta, 1996; Wesley-Hunt, 2005) using data reanalyzed from Legendre and Roth (1988). These produced reasonable estimates across the entire observed size range for both clades. For several amphicyonid taxa, only lower carnassial length measurements could be obtained. In those cases, body masses were estimated using a regression of lower carnassial length across all caniform taxa, reanalyzed from data in Van Valkenburgh (1990). Regression equations and coefficients are given in Table 2. Additional body mass estimates for fossil caniforms were acquired from the Neogene of the Old World database (NOW; 2003) and Viranta (1996). Body mass data are summarized in Appendix 1 (available at the Society of Systematic Biologists website, http://systematicbiology.org).
|
This study also incorporated temporal data into calculations for reconstructing ancestral states, in the form of variable branch lengths. Because temporal differences in the appearances of taxa represent significant amounts of time relative to evolutionary rates, the date of the "tip" must be taken into consideration (Drummond et al., 2001, 2002; Rambaut, 2000). Temporal data in this study constrain not only internal branch lengths, but also lengths and termination points for terminal branches (Fig. 2), with branch lengths representing millions of years. First appearances were taken as the "tip" dates for all terminal taxa. Dates of divergence between clades were then established using the earliest appearing taxon within a monophyletic clade (Fig. 2, inset). Lengths of terminal branches from the nearest node are measured by the difference in time from the first appearance of a given taxon and the first appearance of the earliest taxon in that clade.
By applying tip dates, variation is created in the tip-to-root distance across taxa. This approach assumes that taxa closer to a node have diverged less from the nodal character state than have more distant taxa (Pagel, 1999; 2002; Webster and Purvis, 2002b). Temporal data were compiled from the literature, the NAFMSD (Alroy, 2002) and NOW (2003) databases (summarized in Appendix 1, available online at http://systematicbiology.org).
Ancestral State Reconstruction
Ancestral state reconstructions were executed in the Mesquite software package (Maddison and Maddison, 2004; Maddison et al., 2002), for log-transformed body mass data using weighted squared-change parsimony (WSP; Maddison, 1991). WSP minimizes the sum of squared change along all branches of the tree, weighting branches by their length, and is equivalent to the ML estimate assuming a Brownian motion model of evolution (see Polly, 2001; Webster and Purvis, 2002a). Standard errors of the estimated ancestral body sizes were calculated using the PDAP module in Mesquite (Midford et al., 2003), following the methodology described in Garland and Ives (2000) and employed by Laurin (2004) to create 95% confidence intervals (CIs) around these estimates. As the framework phylogenetic tree contains "soft" polytomies, representing ambiguity regarding true branching order rather than hypotheses of zero-length branches, N-2-z degrees of freedom were applied (where N is the number of taxa, and z the number of polytomies), conservatively expanding CIs around the estimate (Garland and Diaz-Uriarte, 1999). High and low CI values and the estimated node values were calculated in log-transformed units and then converted into kilogram body masses to avoid issues with confidence interval de-transformation biases (see Smith, 1993).
Ancestral body sizes were reconstructed for four hypothetical ancestors. Because of the phylogenetic uncertainty noted above, node-based definitions were applied for (1) Caniformia (the last common ancestor of all extant caniform taxa); (2) Arctoidea (the last common ancestor of Musteloidea, Pinnipedia, and Ursidae); (3) the unnamed node representing the last common ancestor of musteloids and pinnipeds; and (4) Musteloidea (the last common ancestor of Ailurus, mephitids, mustelids [sensu stricto] and procyonids) (see Fig. 1). The inclusion of various fossil groups in each of these clades necessarily changed, depending on the topology being evaluated. For example, if Amphicyonidae is considered the sister clade to Ursidae, then by the above definitions they are included within both the Caniformia and Arctoidea. If they are considered the sister to Canidae, they are excluded from the Arctoidea, but are still considered part of the Caniformia (see Fig. 3).
WSP can incorporate branch lengths into its calculations. As such, the effects of incorporating temporal data into the reconstruction of ancestral body size in the Caniformia can be assessed. A reanalysis of the all-taxa data set was performed, discarding temporal information by setting all branch lengths equal to 1. Setting all branch lengths equal to one another assesses the validity of the assumption that all of the necessary information for reconstructing the evolution of a character is contained solely in the character state distribution among terminal taxa and the branching order of the cladogram.
|
Results |
|---|
|
TopAbstractBackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
The WSP reconstructions of ancestral body size within the Caniformia demonstrate that incorporating fossil data into reconstructions generally reduces ambiguity around estimates, and overturned several patterns that were recovered when only the extant taxon subset was analyzed. These results document both the complexity in body size evolution among caniform carnivorans, and a more general need to combine fossil data within a robust phylogenetic framework for interpreting character evolution, where possible.
The Impact of Incorporating Fossil Taxa on Body Size Reconstructions
A comparison of reconstructed values at each of the four ancestral nodes, using extant-only and all-taxa (extant plus fossil) data sets, reveals dramatically different estimates of ancestral body size. The WSP reconstructions of body mass for the Caniformia and Arctoidea are both larger for the extant-only data set than for the all-taxa data set (Table 3, A, B). Body size estimates are less than 5 kg for both nodes when fossil taxa are included in the reconstruction, while the extant-only set reconstructs estimates slightly greater than 20 kg, with complete separation of the 95% CI intervals for the estimated node values (Table 3, A, B). It should also be noted that for each of three nodes (Caniformia, Arctoidea, and Musteloidea) standard errors for the reconstructions are substantially smaller when fossil data are included.
|
The unnamed node uniting Pinnipedia and Musteloidea deserves particular mention. While the WSP estimate for the extant-only set is larger than that produced by the all-taxa set, the extant-only CIs are completely subsumed by those of the all-taxa set (Table 3, A). In fact, the CIs around the estimated body size for the all-taxa set are completely uninformative, ranging from slightly more than a kilogram to about 180 kg. These large errors are due to the fact that there are currently no reliable body size estimates for fossil pinnipeds. Deriving a reasonable body size proxy that can be extrapolated to fossil taxa in this clade has proven difficult, as pinnipeds feed in different ways than terrestrial carnivorans and do not routinely support their body weight on their limbs. Therefore, many of the commonly used morphological proxies for body size estimation (dental and limb bone measures) are unlikely to provide accurate estimates. In addition, due to the thermal constraints of an aquatic habitat, pinnipeds add mass to their bodies in ways that are not analogous to terrestrial mammals, and the rules governing relationships of body mass to morphologic measurements probably are significantly different. Fossil pinnipeds were excluded from the analyses, although temporal information from first appearances (Démére et al., 2003) were used to calibrate the divergence ages of the major family and subfamily groups. The result is a phylogenetic tree with long branches from the node defining the LCA of pinnipeds and musteloids to the extant pinnipeds, and because pinnipeds exclusively occupy the large end of the modern body size distribution, these long branches lead to large errors around the reconstructed values at the ancestral node. This underscores the importance of incorporating fossil data to constrain reconstructions.
However, it is possible that the small-bodied reconstructions for the LCAs of the Caniformia, Arctoidea, and Musteloidea clades are the result of excluding potentially large fossil pinnipeds from the analysis. To test the sensitivity of the reconstructions to these missing data, the fossil pinnipedimorph taxa documented in Démére et al. (2003) were incorporated into the phylogenetic tree, but were all assigned a mass of 1600 kg, equivalent to the largest living caniform species: Miroungaleonina, the southern elephant seal. This is unrealistically large for most, if not all, fossil pinnipeds, and represents an extreme test of the validity of the small-bodied ancestors reconstructed for the caniform and arctoid nodes. Reanalysis with all fossil pinnipeds having elephant seal body masses does not alter the reconstructions for the LCAs of the Caniformia, Arctoidea, or Musteloidea (Table 3, C), and demonstrates that reconstructed body sizes less than 5 kg for the Caniformia and Arctoidea in particular are robust even to what are almost certainly drastic overestimates of fossil pinniped body sizes. As it is highly likely that many of these taxa had body masses considerably less than 1600 kg, reconstructed body sizes of less than 5 kg for the LCAs of Caniformia and Arctoidea can be made with confidence.
The Impact of Incorporating Temporal Data on Body Size Reconstructions
The previous analysis included variable branch lengths obtained from the first appearances to calibrate clade divergences and tip dates for both extant-only and all-taxa data sets. A reanalysis of the all-taxa data set was performed, excluding temporal information by setting all branch lengths equal to 1. For the Caniformia, Arctoidea, and Pinnipedia + Musteloidea, both the estimates and large CIs around the estimates for the reconstruction excluding temporal information, while slightly smaller for the Caniformia and Arctoidea reconstructions, show a similar pattern to the extant-only analysis (Table 3, D). The reconstructions for these nodes are near the middle of the broad observed range, and estimated size is somewhat larger, indicating that early-appearing taxa in these clades are smaller than those appearing later. This pattern is reversed in the Musteloidea, with a dramatic reduction in the reconstructed size. Reconstructions were also accompanied by increased ambiguity in the reconstructions (larger CIs around the estimate) for each of the three nodes when the temporal data were discarded. These patterns document the presence of diverse ranges of body sizes within the major lineages, although the difference in the reconstructions when temporal data are included in branch lengths indicates that the distribution of the observed body sizes through time is not uniform in these clades. Therefore, temporal data, incorporated into these analyses as branch lengths, are necessary for accurate body size reconstruction.
The Sensitivity of Body Size Reconstruction to Uncertainty in Phylogeny
The previous analyses focused on reconstructions of ancestral body size for a single topology (Fig. 2): Amphicyonidae as the sister to all extant caniforms (Wesley-Hunt and Flynn, 2005) and "Paleomustelidae" considered as a stem lineage basal to mephitids, procyonids, and mustelids (i.e., as stem musteloids) (Baskin, 1998). As noted above, it is not certain that these are the correct positions for these groups. The analyses were repeated for each of the eight alternative phylogenies for the positions of Amphicyonidae (four topologies) and "Paleomustelidae" (two topologies) (Fig. 3). Character state reconstructions were generally unaffected by varying the phylogenetic positions of either amphicyonids or "paleomustelids." A comparison of reconstructions relative to the four alternative positions for Amphicyonidae are presented for two representative nodes: Arctoidea and Musteloidea (Table 4). Reconstructions incorporating the alternative positions for "Paleomustelidae" are shown for all four nodes: Caniformia, Arctoidea, Pinnipedia + Musteloidea, and Musteloidea (Table 3, E).
|
Reconstructions for the four alternative positions of Amphicyonidae show no significant changes in the reconstructed values for the Arctoidea or Musteloidea nodes (Table 4). Values for the WSP reconstruction of Musteloidea ancestral body mass are nearly identical in each amphicyonid topology. For Arctoidea, values are similar except for the topology uniting Ursidae and Amphicyonidae as sister taxa, for which the estimated body size is somewhat larger, although the CIs for all of the reconstructions overlap (Table 4). As with the amphicyonids, varying the position of "Paleomustelidae" between alternate topologies had no effect on reconstructed body sizes (Table 3, A, E).
|
Discussion |
|---|
|
TopAbstractBackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
Ancestral Body Size in the Caniformia—The Importance of Fossil Data
Body mass estimates for fossil taxa show larger ranges in body size than are observed among the extant taxa in each of the major caniform lineages, especially among canids and ursids. It was noted above that fossil ursids span a much larger range of absolute body size than do extant ursids. Furthermore, comparing the range of body sizes at the beginning and end of the ursid lineage history (e.g., the first and last 10 million years of ursid evolution: roughly 38 to 28 Mya, and 10 Mya to Recent) shows a marked difference in the size range of early bears compared to later taxa (Table 5). When fossil taxa are excluded from the analysis, the greater range of body sizes for this clade is lost; when the accompanying temporal data are discarded, the trend of increasing body size is lost.
|
In several studies it has been noted that the confidence intervals around root node reconstructions can often exceed the range of observed tip values (Garland et al., 1999; Polly, 2001; Schluter et al., 1997). However, CIs around root node estimates in the present study are narrowed substantially when the fossil data are included. This is a result not only of the additional character state observations, but more importantly, of the associated temporal information. Taxa near estimated nodes contribute significantly less variance to the reconstructed value under a Brownian motion model (Felsenstein, 1985). Indeed the opposite pattern was already noted in the large CIs around the Pinnipedia + Musteloidea node due to the long branches that result from missing body mass data for pinnipeds. Thus, the addition of taxa at or near ancestral nodes serves to inform the value of the estimate, and also to constrain CIs around that estimate (Garland et al., 1999).
Polly (2001) calculated a 95% confidence envelope for Paleogene carnivoramorphan body mass estimates. These represented the range of values spanning 95% of possible end values for a Brownian motion model given an evolutionary rate, an initial body mass, and an observed amount of elapsed time. He found that the confidence envelope spanned a far greater range of body size than all known extant animals, which would seem to discount the utility of Brownian motion models in ancestral state reconstruction. In fact, a similar confidence envelope can be calculated for the all-taxa analysis in this study. Using the estimated evolutionary rate as the variance parameter and the reconstructed ancestral body mass of the Caniformia node as the initial state of a Brownian motion model (0.73 and 1.72 kg, respectively; Table 3, A) spanning 38 million years (from the FAE of Parictis parvus), the upper and lower 95% confidence envelope values exceed the observed range of caniform body masses by a considerable margin (Table 6).
|
However, Polly's (2001) example assumed that a Brownian motion model proceeded at a given rate for several million generations to a set of coeval tip observations (Polly, 2001:606). In Table 6, a similar assumption is made, where the observed range of caniform body sizes is assumed to exist 38 million years after the existence of a 1.72-kg ancestor, given an estimated per-million-year evolutionary rate. But the tree in the all-taxa analysis is not ultrametric; the caniform taxa are arrayed through time relative to their appearance in the fossil record, and the distances between the reconstructed ancestor at the Caniformia node and the observed tip values for most taxa are considerably less than 38 million years. Thus, a simple Brownian motion model is not appropriate in this case.
The temporal data associated with the appearance of large body sizes within the Ursidae, Amphicyonidae, and Canidae play a significant role in reconstructing small ancestral size for both the Caniformia and Arctoidea, and demonstrate the general importance of branch length information in ancestral character state reconstructions (Garland et al., 1999; Polly, 2001). When the temporal data adjust branch lengths, the impact of each character state observation on the reconstruction is weighted according to its temporal proximity to that node, thereby incorporating evolutionary trends in the character of interest into the reconstruction of the ancestral state.
This may seem odd, in that previous studies of reconstruction methods such as weighted squared-change parsimony have shown that directional trends can be a serious set-back to accurate reconstructions (Oakley and Cunningham, 2000; Webster and Purvis, 2002b). Both the extant-only and uniform branch length analyses in this study support these previous observations. If taxa early in the evolutionary history of a clade are small, whereas later taxa are large (as with bears in the present study), then a reconstructed ancestral size for the LCA using either the extant taxa-only set or equal branch lengths cannot return accurate estimates. However, including data for fossil taxa serves to add character information that is not observed in the modern taxa and also weights the estimate of the nodal reconstruction (Garland et al., 1999). Oakley and Cunningham (2000) noted this potentially significant effect in their reconstruction of ancestral character states when adding even a single known ancestor at the root of the phylogeny. Although true ancestors will rarely be known, our analyses document that adding large numbers of fossil taxa that bracket multiple nodes deep within a phylogeny can minimize the biases caused by directional trends.
After completion of these analyses, Wang et al. (2005) published cranial descriptions of three arctoid taxa from Mongolia, providing dental measurements that could be used to estimate body size (Table 7). These taxa (Amphicynodonteilhardi, Amphicticepsdorog, and Amphicticepsshackelfordi) have been placed as basal ursids (see Hunt, 1998b; Wang et al., 2005). Estimated body masses for these newly reported taxa (1.3 to 4.6 kg) are exactly what would be predicted for basal members of the ursid lineage, close to the initial arctoid radiation, in the all-taxa, temporally constrained analyses.
|
These reconstructions demonstrate that many of the taxa that appear early in the lineage histories of several caniform clades (e.g., the Ursidae, Amphicyonidae, and both the Hesperocyoninae and Borophaginae subfamilies within the Canidae) occupy the small end of the size distribution for their respective clades, whereas larger members of these clades generally occur later. As a result, both the phylogenetic position of a taxon of a certain body size and the time when it appears in the history of a clade are crucially important for the reconstruction of patterns and trends in body size evolution. Incorporating fossils therefore yields more precise, and likely more accurate, reconstructions despite the existence of trends across the phylogeny.
The LCA of both the Caniformia and Arctoidea therefore most likely were small-bodied (on the order of 1 to 5 kg), and not large-bodied (on the order of 10 to 50 kg). These differences, nearly an order of magnitude, should have a profound impact on interpretations of the life history, paleoecological interactions, and adaptation to past environments for ancestral caniform and arctoid forms (Gittleman, 1986b, 1991, 1993; Gittleman and Harvey, 1982; Meiri et al., 2004a, 2004b). Qualitatively, when considering an ancestral caniform or arctoid, the extant-only reconstruction of body masses on the order of tens of kilograms (the size of the gray wolf, Canis lupus), as compared to the all-taxa reconstruction of less than 5 kilograms (the size of the striped skunk, Mephitis mephitis) produces a dramatically different view of the biology of the ancestral organism. Quantitatively, body size in living carnivorans, and mammals in general, can be correlated to many life history and ecological variables; for example, average home range size, which correlates strongly with body size in Carnivora, after correcting for phylogeny (Gittleman, 1986b, 1993; Gittleman and Harvey, 1982; but see Kelt and Van Vuren, 2001). Unfortunately, most ecological variables do not readily preserve in the fossil record. However, characters such as body mass can be estimated for fossil taxa, and with more precise and accurate estimates of ancestral body masses, it is possible to reconstruct a more reliable picture of the life history of ancestral caniforms. These results illustrate the more general potential of studies intending to secure accurate morphologic correlates for a broad range of organismal attributes having low preservation rates (e.g., Gittleman, 1986a, 1986b, 1993; Gittleman and Harvey, 1982).
Small ancestral body size reconstructions, when incorporating fossil data, document parallel, independent trends of increased body size within the canid, ursid, and amphicyonid lineages. The small to medium body size reconstruction for the LCA of Pinnipedia + Musteloidea potentially implies another independent acquisition of large body size along the pinniped lineage, although accurate body size proxies for fossil members of this group will be necessary to verify this with any certainty. These results confirm the conclusions of several prior studies of body size evolution within less inclusive caniform groups, which documented general size increases in two subfamilies of Canidae (Van Valkenburgh et al., 2004), the Ursidae (Hunt, 1998b), and the Amphicyonidae (Hunt, 1998a), and are consistent with the so-called "Cope's rule" of size increase observed in some lineages through time (Alroy, 1998; Stanley, 1973).
Ancestral reconstructions for the Musteloidea document the opposite pattern. If only extant taxa are included, a 5-kg ancestor is reconstructed, albeit with large errors. However, a 10-to 20-kg ancestor is recovered when the fossil taxa are included. This demonstrates that, as with canids and ursids, the distribution of extant musteloid body sizes does not fully document the body size distribution of this clade throughout its evolutionary history. In contrast, modern musteloids fail to represent the larger-bodied taxa found in their fossil record. The ancestral musteloid body size estimate of 15.75 kg (Table 3, A) corresponds to the 93rd and 86th percentiles for extant musteloids and all musteloids, respectively. Both median body size (extant-only: 1.45 kg, all-taxa: 3.2 kg) and maximum body size (extant-only: 30.6 kg, all-taxa: 100 kg) are lower for the Musteloidea when fossil taxa are excluded, whereas minimum body size remains the same (104 g). Thus, minimum body size for this clade is observed in the Recent, average body size is smaller now than for the entire lineage history, and maximum body size for this clade was observed in a fossil taxon, documenting size decrease among musteloids through time. Although "Cope's rule" is a repeated pattern among at least three terrestrial caniform lineages, it is not a pervasive rule of body size evolution throughout the Caniformia.
Beyond clade-specific results for the Caniformia, the impact that fossil data have on body size reconstruction in this case serves to emphasize the general importance of incorporating character state observations and temporal data from the fossil record. The present data set documents several independent trends of body size evolution. In addition, several lineages (most notably Canidae and Ursidae) possess fossil taxa that far exceed the upper limits of the modern body size distribution. For both of these clades, early appearing members fall below the lower size limit of the range observed in Recent taxa. Directional trends and phenotypic values outside of the modern observed range are generally regarded as "fatal" to ancestral state reconstructions (Garland et al., 1999). To the contrary, if sufficient information is available for a character in the fossil record of a group, then trends do not necessarily pose the same serious problem. In fact, the addition of character state observations among fossil members of a clade, and the timing of their appearance, help to constrain the range of potentially viable hypotheses of ancestral character states and provide the only way to actually document such evolutionary trends. As such, observed patterns of character states and their distribution through time in fossil taxa should be incorporated into reconstructions in cases where the information is preserved in the fossil record, and the fossil taxa can be reliably placed into a phylogenetic framework.
|
Conclusions |
|---|
|
TopAbstractBackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
Most recent phylogenies place extant caniform carnivorans into four main lineages: Canidae, Ursidae, Pinnipedia, and Musteloidea (the latter three comprising a monophyletic clade: Arctoidea). The distribution of body sizes of extant caniforms appears to represent a pattern of decreasing body size across the phylogeny, wherein large body sizes are observed in the ursids and pinnipeds, while musteloids are almost exclusively small-to medium-bodied. However, estimated body sizes for many fossil canids, ursids, and musteloids taxa fall well outside of the ranges observed for living taxa, demonstrating that the modern distribution of caniform body sizes (with many lineages presently represented almost exclusively by large-or small-bodied forms) is not representative of the entire history of the clade.
Incorporating fossil taxa into the reconstructions of the body sizes for the last common ancestors of the Caniformia and the Arctoidea points to a small-bodied ancestor, between 1 and 5 kg, for both clades. In addition, it is likely that the last common ancestor of musteloids was a small-to medium-sized animal (between 10 and 20 kg). Therefore, analyses of body size evolution in the Caniformia that do not incorporate data from fossil taxa are hindered because much of the evolutionary history of this character is not represented in the modern subsample of taxa. Due to a lack of reliable body mass estimates for fossil pinnipeds, the errors associated with the reconstructions of the LCA for Pinnipedia + Musteloidea are extreme. Data for fossil taxa will be necessary to make a definitive statement about body size evolution in the Pinnipedia.
The small body sizes reconstructed for the most basal divergences in the Caniformia, when fossils are included in the analyses, reveal independent increases in body size along at least three lineages of terrestrial caniforms (Canidae, Ursidae, and Amphicyonidae), consistent with previous observations within these groups (Hunt, 1998a, 1998b; Van Valkenburgh et al., 2004; Wang, 1994; Wang et al., 1999). More importantly, the persistence of small ancestral body size reconstructions for the LCAs of Caniformia and Arctoidea, irrespective of the position of amphicyonids, indicates that the acquisition of large body sizes in each of these groups was achieved in parallel. In a contrasting pattern, body size reconstructions for the LCA of Musteloidea shows evidence for body size decrease. Thus, "Cope's rule" may be a recurring, but not a universal, theme in the evolution of body size in the Caniformia.
More generally, these analyses demonstrate the importance of incorporating all available data, including character observations and temporal information for fossil taxa, into the reconstruction of ancestral character states. Incorporating fossil data improved weighted squared-change parsimony reconstructions. Additional character state observations in fossil taxa serve to increase the precision of ancestral character state reconstructions and better constrain potential hypotheses of character evolution by offering a more complete knowledge of the distribution of character states in the clade under study. This can provide evidence to falsify or more robustly support hypotheses based solely on the distribution of character observations among extant taxa. Temporal information incorporated into first appearance ages and branch lengths alters the impact that observed character states have on the reconstructed values of a particular node, and incorporates evidence of directional trends into the reconstruction of ancestral states. These results further emphasize the importance of finding accurate morphological correlates for characters of interest that do not readily preserve in the fossil record (e.g., Gittleman, 1986a, 1986b, 1993; Gittleman and Harvey, 1982; Lindenfors et al., 2003) and the potentially beneficial impact that incorporating these correlates as proxies can have in character state reconstruction.
|
Acknowledgements |
|---|
We thank A. Goswami, L. Van Valen, P. Wagner, and the students of the Columbia University Vertebrate Paleontology Seminar for helpful insights during the course of this study and comments on manuscript drafts. M. Butler and P. D. Polly provided helpful and constructive reviews of the manuscript, for which we are appreciative. This research was partially supported by the Field Museum of Natural History and National Science Foundation grants to J. J. F., M. Nedbal (DEB-9707225), A. Yoder, J. J. F., and M. Nedbal (DEB-9807045). DNA sequencing for the framework molecular phylogeny of Flynn et al. (2005) was performed at the Field Museum's Pritzker Laboratory for Molecular Systematics and Evolution, operated through the generous support from The Pritzker Foundation.
|
References |
|---|
|
TopAbstractBackgroundMethodsResultsDiscussionConclusionsReferences |
|---|
-
Alroy J. Cope's rule and the dynamics of body mass evolution in North American fossil mammals. Science (1998) 280:731–734.
Alroy J. North American fossil mammal systematics database. (2002) http://www.nceas.ucsb.edu/~alroy/nafmsd.html.
Baskin J. A. Mustelidae. In: Evolution of tertiary mammals of North America—Janis C. M., Scott K. M., Jacobs L. L., eds. (1998) New York: Cambridge University Press. Pages 152–173.
Bininda-Emonds O. R. P., Gittleman J. L., Purvis A. Building large trees by combining phylogenetic information: A complete phylogeny of the extant Carnivora (Mammalia). Biol. Rev. (1999) 74:143–175.
Cunningham C. W., Omland K. E., Oakley T. Reconstructing ancestral states: A critical reappraisal. Trends Ecol. Evol. (1998) 13:361–366.[CrossRef]
Davis C. S., Delisle I., Stirling I., Siniff D. B., Strobeck C. A phylogeny of the extant Phocidae inferred from complete mitochondrial DNA coding regions. Mol. Phylogenet. Evol. (2004) 33:363–377.[CrossRef][Web of Science][Medline]
Démére T. A., Berta A., Adam P. J. Pinnipedimorph evolutionary biogeography. Bull. Am. Mus. Nat. Hist. (2003) 279:32–76.[CrossRef]
Dragoo J. W., Honeycutt R. L. Systematics of mustelid-like carnivores. J. Mammal. (1997) 78:426–443.[CrossRef][Web of Science]
Drummond A. J., Forsberg R., Rodrigo A. G. The inference of stepwise changes in substitution rates using serial sequence samples. Mol. Biol. Evol. (2001) 18:1365–1371.
Drummond A. J., Nicholls G. K., Forsberg R., Solomon W. Estimating mutation parameters, population history and genealogy simultaneously from temporally spaced sequence data. Genetics (2002) 161:1307–1320.
Felsenstein J. Phylogenies and the comparative method. Am. Nat. (1985) 125:1–15.[CrossRef][Web of Science]
Flynn J. J., Finarelli J. A., Zehr S., Hsu J., Nedbal M. A. Molecular phylogeny of the Carnivora (Mammalia): Assessing the impact of increased sampling on resolving enigmatic relationships. Syst. Biol. (2005) 54:317–337.
Flynn J. J., Galiano H. Phylogeny of Early Tertiary Carnivora, with a description of a new species of Protictis from the Middle Eocene of Northwestern Wyoming. Am. Mus. Novitates (1982) 2725:1–64.
Flynn J. J., Nedbal M. A. Phylogeny of the Carnivora (Mammalia): Congruence vs. incompatibility among multiple data sets. Mol. Phylogenet. Evol. (1998) 9:414–426.
Flynn J. J., Nedbal M. A., Dragoo J. W., Honeycutt R. L. Whence the red panda? Mol. Phylogenet. Evol. (2000) 17:190–199.[CrossRef][Web of Science][Medline]
Flynn J. J., Wesley-Hunt G. D. Carnivora. In: Origin, timing, and relationships of the major clades of extant placental mammals—Archibald D., Rose K. D., eds. (2005) Baltimore, Maryland: Johns Hopkins University Press. Pages 175–198.
Garland T., Diaz-Uriarte R. Polytomies and phylogenetically independent contrasts: Examination of the bounded degrees of freedom approach. Syst. Biol. (1999) 48:547–558.
Garland T., Ives A. R. Using the past to predict the present: Confidence intervals for regression equations in phylogenetic comparative methods. Am. Nat. (2000) 155:346–364.[CrossRef][Medline]
Garland T., Midford P. E., Ives A. R. An introduction to phylogenetically based statistical methods, with a new method for confidence intervals on ancestral values. Am. Zool. (1999) 39:374–388.[Web of Science]
Gaubert P., Veron G. Exhaustive sample set among Viverridae reveals the sister-group of felids: The linsangs as a case of extreme morphological convergence within Feliformia. Proc. R. Soc. Lond. B Biol. Sci. (2003) 270:2523–2530.[CrossRef][Medline]
Ginsburg L. Les amphicyons des Phosporites du Quercy. Ann. Paléontol. (Vert.) (1966) 52:21–64.
Gittleman J. L. Carnivore brain size, behavioral ecology, and phylogeny. J. Mammal. (1986a) 67:23–36.[CrossRef][Web of Science]
Gittleman J. L. Carnivore life history patterns: Allometric, phylogenetic, and ecological associations. Am. Nat. (1986b) 127:744–771.[CrossRef][Web of Science]
Gittleman J. L. Carnivore olfactory bulb size: Allometry, phylogeny and ecology. J. Zool. (1991) 225:253–272.[CrossRef][Web of Science]
Gittleman J. L. Carnivore life histories: A reanalysis in light of new models. Dunstone N., Gorman M., eds. (1993) Mammals as predators: The proceedings of a symposium held by the Zoological Society of London and the Mammal Society: London. New York: Clarendon Press. Pages 65–86.
Gittleman J. L., Harvey P. H. Carnivore home-range size, metabolic needs and ecology. Behav. Ecol. Sociobiol. (1982) 10:57–63.[CrossRef][Web of Science]
Gittleman J. L., Purvis A. Body size and species-richness in carnivores and primates. Proc. R. Soc. Lond. B Biol. Sci. (1998) 265:113–119.[Medline]
Gittleman J. L., Van Valkenburgh B. Sexual dimorphism in the canines and skulls of carnivores: Effects of size, phylogeny, and behavioral ecology. J. Zool. (1997) 242:97–117.[CrossRef][Web of Science]
Harvey P. H., May R. M., Nee S. Phylogenies without fossils. Evolution (1994) 48:523–529.[CrossRef][Web of Science]
Hough J. R. The auditory region in some members of the Procyonidae, Canidae and Ursidae: Its significane in the phylogeny of the Carnivora. Bull. Am. Mus. Nat. Hist. (1948) 92:67–118.
Hunt R. M. The auditory bulla in Carnivora: An anatomical basis for reappraisal of carnivore evolution. J. Morphol. (1974) 1143:21–76.
Hunt R. M. Basicranial anatomy of Cynelos Jourdan (Mammalia: Carnivora), an Aquitanian Amphicyonid from the Allier Basin, France. J. Paleontol. (1977) 51:826–243.[Abstract]
Hunt R. M. Amphicyonidae. In: The terrestrial Eocene-Oligocene transition in North America—Prothero D. R., Emry R. J., eds. (1996) New York, New York: Cambridge University Press. Pages 476–485.
Hunt R. M. Amphicyondae. In: Evolution of tertiary mammals of North America—Janis C. M., Scott K. M., Jacobs L. L., eds. (1998a) New York: Cambridge University Press. Pages 196–227.
Hunt R. M. Ursidae. In: Evolution of tertiary mammals of North America—Janis C. M., Scott K. M., Jacobs L. L., eds. (1998b) New York: Cambridge University Press. Pages 174–195.
Kelt D. A., Van Vuren D. H. The ecology and macroecology of mammalian home range size. Am. Nat. (2001) 157:637–645.[CrossRef][Web of Science][Medline]
Koepfli K. P., Wayne R. K. Phylogenetic relationships of otters (Carnivora: Mustelidae) based on mitochondrial cytochrome b sequences. J. Zool. (1998) 246:401–416.[CrossRef][Web of Science]
Koepfli K. P., Wayne R. K. Type ISTS markers are more informative than cytochrome b in phylogenetic reconstruction of the Mustelidae (Mammalia: Carnivora). Syst. Biol. (2003) 52:571–593.
Kozlowski J., Weiner J. Interspecific allometries are by-products of body size optimization. Am. Nat. (1997) 149:352–380.[CrossRef][Web of Science]
Laurin M. The evolution of body size, Cope's rule and the origin of amniotes. Syst. Biol. (2004) 53:594–622.
Ledje C., Arnason U. Phylogenetic analyses of complete cytochrome b genes of the order Carnivora with particular emphasis on the Caniformia. J. Mol. Evol. (1996a) 42:135–144.[CrossRef][Web of Science][Medline]
Ledje C., Arnason U. Phylogenetic relationships within caniform carnivores based on analyses of the mitochondrial 12S rRNA gene. J. Mol. Evol. (1996b) 43:641–649.[CrossRef][Web of Science][Medline]
Legendre S., Roth C. Correlation of carnassial tooth size and body weight in Recent carnivores (Mammalia). Hist. Biol. (1988) 1:85–98.[CrossRef]
Lindenfors P., Dalen L., Angerbjorn A. The monophyletic origin of delayed implantation in carnivores and its implications. Evolution (2003) 57:1952–1956.[Web of Science][Medline]
Maddison W. P. Squared-change parsimony reconstructions of ancestral statesfor continuous-valued characters on a phylogenetic tree. Syst. Zool. (1991) 40:304–314.[Abstract]
Maddison W. P. Calculating the probability distributions of ancestral states reconstructed by parsimony on phylogenetic trees. Syst. Biol. (1995) 44:474–481.
Maddison W. P., Maddison D. R. Mesquite: A modular system for evolutionary analysis, version 1.05. (2004) http://mesquiteproject.org.
Maddison W. P., Maddison D. R., Midford P. TreeFarm: A package of modules for Mesquite, version 1.0. (2002).
Martins E. P. Estimation of ancestral states of continuous characters: A computer simulation study. Syst. Biol. (1999) 48:642–640.
Matthew W. D. Third contribution to the Snake Creek Fauna. Bull. Am. Mus. Nat. Hist. (1924) 50:59–210.
Meiri S., Dayan T., Simberloff D. Body size of insular Carnivores: Little Support for the Island Rule. Am. Nat. (2004a) 163:469–479.[CrossRef][Web of Science][Medline]
Meiri S., Dayan T., Simberloff D. Carnivores, biases and Bergmann's rule. Biol. J. Linn. Soc. (2004b) 81:579–588.[CrossRef][Web of Science]
Midford P., Garland T., Maddison W. P. PDAP package (of Mesquite). (2003).
NOW database. (2003) http://www.helsinki.fi/science/now/index.html. version July 17, 2003.
Oakley T. H., Cunningham C. W. Independent contrasts succeed where ancestor reconstruction fails in a known bacteriophage phylogeny. Evolution (2000) 54:397–405.[CrossRef][Web of Science][Medline]
Pagel M. Inferring the historical patterns of biological evolution. Nature (1999) 401:877–884.[CrossRef]
Pagel M. Modelling the evolution of continuously varying characters: The case of hominid cranial capacity. In: Morphology, shape and phylogeny—McLeod N., Forey P. L., eds. (2002) New York: Taylor & Francis. Pages 269–286.
Pagel M., Meade A., Barker D. Bayesian rstimation of ancestral character states on phylogenies. Syst. Biol. (2004) 53:673–684.
Polly P. D. Paleontology and the comparative method: Ancestral node reconstructins versus observed node values. Am. Nat. (2001) 157:596–609.[CrossRef][Web of Science][Medline]
Purvis A., Gittleman J. L., Luh H. K. Truth or consequences—effects of phylogenetic accuracy on 2 comparative methods. J. Theor. Biol. (1994) 167:293–300.[CrossRef][Web of Science]
Rambaut A. Estimating the rate of molecular evolution: Incorporating noncontemporaneous sequences into maximum likelihood phylogenies. Bioinformatics (2000) 16:395–399.
Salisbury B. A., Kim J. H. Ancestral state estimation and taxon sampling density. Syst. Biol. (2001) 50:557–564.
Sato J. J., Hosoda T., Wolsan M., Suzuki H. Molecular phylogeny of arctoids (Mammalia: Carnivora) with emphasis on phylogenetic and taxonomic positions of the ferret-badgers and skunks. Zool. Sci. (2004) 21:111–118.[CrossRef][Web of Science][Medline]
Sato J. J., Hosoda T., Wolsan M., Tsuchiya K., Yamamo M., Suzuki H. Phylogenetic relationships and divergence times among mustelids (Mammalia: Carnivora) based on nucleotide sequences of the nuclear interphotoreceptor retinoid binding protein and mitochondrial cytochrome b genes. Zool. Sci. (2003) 20:243–264.[CrossRef][Web of Science][Medline]
Schluter D., Price T., Mooers A. O., Ludwig D. Likelihood of ancestor states in adaptive radiation. Evolution (1997) 51:1699–1711.[CrossRef][Web of Science]
Schmidt-Nielsen K. Scaling: Why is animal size so important? (1984) Cambridge: Cambridge University Press.
Smith A. B. Systematics and the fossil record (1994) Boston: Blackwell Scientific Publications.
Smith F. A., Lyons S. K., Ernest S. K. M., Jones K. E., Kaufman D. M., Dayan T., Marquet P. A., Brown J. H., Haskell J. P. Body mass of late Quaternary mammals. Ecology (2003) 84:3402.[CrossRef][Web of Science]
Smith R. J. Logarithmic transformation bias in allometry. Am. J. Phys. Anthropol. (1993) 90:215–228.[CrossRef][Web of Science]
Stanley S. M. An explanation for Cope's rule. Evolution (1973) 27:1–26.[CrossRef][Web of Science]
Swofford D. L., Maddison W. P. Reconstructing ancestral character states under Wagner parsimony. Math. Biosci. (1987) 87:199–229.[CrossRef][Web of Science]
Tedford R. H., Taylor B. E., Wang X. M. Phylogeny of the Caninae (Carnivora: Canidae): The living taxa. Am. Mus. Novitates (1995) 3146:1–37.
Van Valkenburgh B. Skeletal and dental predictors of body mass in carnivores. In: Body size in mammalian paleobiology: Estimation and biological implications—Damuth J., MacFadden B. J., eds. (1990) New York: Cambridge University Press. Pages 181–206.
Van Valkenburgh B., Wang X. M., Damuth J. Cope 's Rule, hypercarnivory, and extinction in North American canids. Science (2004) 306:101–104.
Veron G., Colyn M., Dunham A. E., Taylor P., Gaubert P. Molecular systematics and origin of sociality in mongooses (Herpestidae, Carnivora). Mol. Phylogenet. Evol. (2004) 30:582–598.[CrossRef][Web of Science][Medline]
Viranta S. European Miocene Amphicyonidae—taxonomy, systematics and ecology. Acta Zool. Fenni. (1996) 204:1–61.
Wang X. M. Phylogenetic sytematics of the Hesperocyoninae (Carnivora: Canidae). Bull. Am. Mus. Nat. Hist. (1994) 221:1–207.
Wang X. M., McKenna M., Dashzeveg D. Amphicticeps and Amphicynodon (Arctoidea, Carnivora) from Hsanda Gol Formation, Central Mongolia and phylogeny of Basal Arctoids with comments on zoogeography. Am. Mus. Novitates (2005) 3483:1–57.[CrossRef]
Wang X. M., Tedford R. H., Taylor B. E. Phylogenetic sytematics of the Borophaginae (Carnivora: Canidae). Bull. Am. Mus. Nat. Hist. (1999) 243:1–391.
Webster A. J., Gittleman J. L., Purvis A. The life history legacy of evolutionary body size change in carnivores. J. Evol. Biol. (2004) 17:396–407.[CrossRef][Web of Science][Medline]
Webster A. J., Purvis A. Ancestral states and evolutionary rates of continuous characters. In: Morphology, shape and Phylogeny—McLeod N., Forey P. L., eds. (2002a) New York: Taylor & Francis. Pages 247–268.
Webster A. J., Purvis A. Testing the accuracy of methods for reconstructing ancestral states of continuous characters. Proc. R. Soc. Lond. B Biol. Sci. (2002b) 269:143–149.[Medline]
Wesley-Hunt G. D. The morphological diversification of carnivores in North America. Paleobiology (2005) 31:35–55.
Wesley-Hunt G. D., Flynn J. J. Phylogeny of the Carnivora: basal relationships among the carnivoramorphans, and assessment of the position of "Miacoidea" relative to crown-clade Carnivora. J. Syst. Palaeontol. (2005) 3:1–28.[CrossRef]
Wozencraft W. C. Order Carnivora. In: Mammal species of the World: A taxonomic and geographic reference—Wilson D. E., Reeder D. M., eds. (1993) Washington, DC: Smithsonian Institution Press. Pages 279–348.
Wyss A. R., Flynn J. J. A phylogenetic analysis and definition of the Carnivora. In: Mammal phylogeny: Placentals—Szalay F., Novacek M., McKenna M., eds. (1993) New York: Springer-Verlag. Pages 32–52.
Yoder A. D., Burns M. M., Zehr S., Delefosse T., Veron G., Goodman S. M., Flynn J. J. Single origin of Malagasy Carnivora from an African ancestor. Nature (2003) 421:734–737.[CrossRef]
Yu L., Li Q., Ryder O. A., Zhang Y. Phylogenetic relationships within mammalian Order Carnivora indicated by sequences of two nuclear DNA genes. Mol. Phylogenet. Evol. (2004a) 33:694–705.[CrossRef][Web of Science][Medline]
Yu L., Li Q., Ryder O. A., Zhang Y. Phylogeny of the bears (Ursidae) based on nuclear and mitochondrial genes. Mol. Phylogenet. Evol. (2004b) 32:480–494.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Jablonski and J. A. Finarelli Congruence of morphologically-defined genera with molecular phylogenies PNAS, May 19, 2009; 106(20): 8262 - 8266. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Li, M. Steel, and L. Zhang More Taxa Are Not Necessarily Better for the Reconstruction of Ancestral Character States Syst Biol, August 1, 2008; 57(4): 647 - 653. [Full Text] [PDF]
|
|
|
|
|
|
M. Hardman and L. M. Hardman The Relative Importance of Body Size and Paleoclimatic Change as Explanatory Variables Influencing Lineage Diversification Rate: An Evolutionary Analysis of Bullhead Catfishes (Siluriformes: Ictaluridae) Syst Biol, February 1, 2008; 57(1): 116 - 130. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Cobbett, M. Wilkinson, and M. A Wills Fossils Impact as Hard as Living Taxa in Parsimony Analyses of Morphology Syst Biol, October 1, 2007; 56(5): 753 - 766. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||