© 2005 Society of Systematic Biologists
Mosaics of Convergences and Noise in Morphological Phylogenies: What's in a Viverrid-Like Carnivoran?
Edited by Olaf Bininda-Emonds
1 Unité Origine, Structure et Evolution de la Biodiversité (CNRS UMR 5202), Département Systématique et Evolution, Muséum National d'Histoire Naturelle, Zoologie: Mammifères et Oiseaux CP 51, 57 rue Cuvier, 75231 Paris Cedex 05, France
2 Division of Natural Sciences, Bethel College 1001 W. McKinley Avenue, Mishawaka, Indiana 46545, USA
3 Current address: Departamento de Biología Aplicada, Estación Biológica de Doñana (CSIC), Avda. María Luisa s/n Pabellón del Perú, 41013 Sevilla, Spain; E-mail: gaubert{at}ebd.csic.es
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
Adaptive convergence in morphological characters has not been thoroughly investigated, and the processes by which phylogenetic relationships may be misled by morphological convergence remains unclear. We undertook a case study on the morphological evolution of viverrid-like feliformians (Nandinia, Cryptoprocta, Fossa, Eupleres, Prionodon) and built the largest morphological matrix concerning the suborder Feliformia to date. A total of 349 characters grouped into four anatomical partitions were used for all species of Viverridae and viverrid-like taxa plus representatives of the Felidae, Hyaenidae, Herpestidae, and one Malagasy mongoose. Recent molecular phylogenetic analyses suggest that viverrid-like morphotypes appeared independently at least three times during feliformian evolution. We thus used a synthetic molecular tree to assess morphological evolutionary patterns characterizing the viverrid-like taxa. We examined phylogenetic signal, convergence and noise in morphological characters using (a) tree-length distribution (g1), (b) partitioned Bremer support, (c) RI values and their distribution, (d) respective contributions of diagnostic synapomorphies at the nodes for each partition, (e) patterns of shared convergences among viverrid-like taxa and other feliformian lineages, (f) tree-length differences among alternative hypotheses, and (g) the successive removal of convergent character states from the original matrix. In addition, the lability of complex morphological structures was assessed by mapping them onto the synthetic molecular tree. The unconstrained morphological analysis yielded phylogenetic groupings that closely reflected traditional classification. The use of a synthetic molecular tree (constraint) combined with our thorough morphological investigations revealed the mosaics of convergences likely to have contributed to part of the historical uncertainty over viverrid classification. It also showed that complex morphological structures could be subjected to reversible evolutionary trends. The morphological matrix proved useful in characterizing several feliformian clades with diagnostic synapomorphies. These results support the removal from the traditionally held Viverridae of several viverrid-like taxa into three distinct families: Nandiniidae (Nandinia), Prionodontidae (Prionodon), and the newly defined Eupleridae (including Cryptoprocta, Fossa, Eupleres plus all "mongoose-like" Malagasy taxa). No clearly "phylogenetically misleading" data subsets could be identified, and the great majority of morphological convergences appeared to be nonadaptive. The multiple approaches used in this study revealed that the most disruptive element with regards to morphological phylogenetic reconstruction was noise, which blured the expression of phylogenetic signal. This study demonstrates the crucial need to consider independent (molecular) phylogenies in order to produce reliable evolutionary hypotheses and should promote a new approach to the definition of morphological characters in mammals.
Keywords: Constrained analysis; convergence; evolutionary scenario; Feliformia; morphology; noise; phylogenetic signal; phylogeny; Viverridae
Received July 3, 2004; Revised October 27, 2004; Accepted March 31, 2005
As recently addressed by Wiens et al. (2003), morphological data often have been considered as highly subjected to adaptive convergence, and thus may yield incorrect reconstructions of evolutionary history (e.g., Hedges and Sibley, 1994; Gatesy et al., 1996; Kusmierski et al., 1997; Alvarez et al., 1999; Teeling et al., 2002). In recent years the use of morphological data has declined with the development of highly efficient sequencing techniques, (see Scotland et al., 2003). This may be due to a decrease in the confidence attributed to morphological characters for phylogenetic reconstruction (or the time associated with the collection and definition of morphological data). The difficulties related to availability and coding of morphological characters include the assessment of homology (e.g., Pimental and Riggins, 1987; Smith, 1994; Wagner, 2001) and the absence of knowledge on models of evolution (Scotland et al., 2003; but see Lewis, 2001). Thus, many papers have focused on the importance of fixing explicit criteria for character selection (Patterson and Johnson, 1997; Hawkins, 2000) in order to base phylogeny on unambiguous characters and to minimize the effect of noise. However, deciphering unambiguous morphological characters ideally would require detailed knowledge of morphogenetic processes for each taxon studied, which is a condition rarely met. Currently most phylogeneticists consider morphological data most appropriately used when compared to molecular phylogenetic data. As a consequence, many empirical and simulation studies have focused on estimating the conflicts between morphological and molecular data sets (e.g., Baker and DeSalle, 1997; Baker et al., 1998; Wiens, 1998; Liu and Miyamoto, 1999; McCracken et al., 1999; Quicke and Belshaw, 1999; Lee, 2001) or between morphological subsets in a molecular phylogeny framework (McCracken et al., 1999; Gaubert et al., 2004a).
The question of how phylogenetic relationships may be affected by morphological convergence remains unclear. In other words, is there a way to quantitatively and qualitatively characterize convergence? A battery of explorative tests are available for estimating phylogenetic signals, levels of homoplasy, and conflicts between data (sub)sets (most of which reviewed in Grant and Kluge, 2003). However, very few studies have focused on character-specific patterns of convergence because investigations have not been explored beyond the partition level. In this context, a first step towards detailed characterization of homoplasy was initiated by Savolainen et al. (2002), who estimated site-by-site "phylogenetic informativeness" through analysis of distribution of Retention Index (RI) values.
Defining Convergence
Convergence may be one of the most cited terms in phylogenetic studies. Surprisingly, clear definitions are noticeably lacking in the most recent papers using this term, and there is little consensus in the literature on what convergence really means (see Wiens et al., 2003). Our aim is not to review here the detailed controversy over the definitions of convergence. We follow a common point of view that convergences are shared derived similarities, which have been acquired independently across lineages during evolution. The distinction from parallelism (i.e., shared derived similarities from a closely related common ancestor) might prove arbitrary and irrelevant in practice because of the difficulty in establishing a clear threshold between closely and not closely related taxa. In addition, several definitions of parallelism are available making the distinction between this term and convergence even more problematic (e.g., Brooks, 1996; Li, 1997; Wiens et al., 2003).
Most tests of evolutionary scenarios have focused on morphological structures related to adaptive functions (e.g., Alvarez et al., 1999; McCracken et al., 1999; Goodacre and Wade, 2001; Parra-Olea and Wake, 2001). However, an important part of the convergence events may not be explainable from an adaptationist point of view (Gould and Lewontin, 1978; Leal et al., 2002). On the other hand, it is also important to consider the fact that "neutral" (i.e., stochastic) morphological convergence might be an artifact due to our own ignorance about the function of particular characters. To disregard this is likely to mislead when trying to establish criteria for detecting erroneous phylogenetic reconstructions (see Wiens et al., 2003). Thus, we will use the term "convergence" to mean independent acquisitions of derived conditions, whether these (1) are clearly of adaptive causes or not and (2) concern character states or complex structures.
Study Group
To further explore the characterization of convergence, we undertook a case study on the morphological evolution of viverrids and viverrid-like feliformians (Mammalia, Carnivora). The suborder Feliformia offers an ideal opportunity to study morphological convergence for two reasons. First, recent works have provided crucial advances in the molecular phylogeny of the group (Veron and Catzeflis, 1993; Flynn and Nedbal, 1998; Veron and Heard, 2000; Gaubert and Veron, 2003; Yoder et al., 2003; Gaubert et al., 2004a, 2004b). This allows for a solid and independent framework to which morphological evolution can be compared. Second, the molecular phylogeny clearly identified three lineages of viverrid-like taxa, implying the independent acquisition of morphological similarities with members of the family Viverridae by the African palm civet (Nandinia), the Malagasy viverrid-like taxa (Eupleres, Fossa, and Cryptoprocta) and the Asiatic linsangs (Prionodon). Such a case of morphotypic convergence is unique within the Feliformia, and may be rare for the whole mammalian order.
The level of similarity between the "true" viverrids (restricted to the subfamilies Hemigalinae, Paradoxurinae and Viverrinae) and the viverrid-like taxa has not been clearly established. Morphological works have focused on "diagnostic" characters such as general shape of the skull, basicranium, dentition, pad morphology, and coat pattern (Ginsburg, 1961; Petter, 1969, 1974; Hunt, 1974, 1987, 2001; Taylor, 1988; Veron, 1999). Viverridae and viverrid-like taxa are distributed throughout tropical Asia and Africa, where they occupy various ranges of habitats, from dense rain forests to woodlands, savannah, and grassland (Kingdon, 1997; Nowak, 1999). They have adapted to almost all possible terrestrial modes of life, from arboreality to strict ground and semiaquatic dwelling (Taylor, 1974, 1976; Veron, 1999). Such a large range of ecomorphs has been considered responsible for the difficulties in reconstructing phylogenetic patterns (Veron, 1995; Gaubert et al., 2002a).
The viverrid-like taxa represent three distinct lineages whose dates of divergence have been estimated by molecular-based methodologies to be between the Oligocene and Miocene (Nandinia: 36 to 40 Mya; Prionodon: 32 to 35 Mya; Malagasy viverrid-like taxa: 18 to 24 Mya [Flynn, 1996; Gaubert and Veron, 2003; Yoder et al., 2003]). The complex mixture of plesiomorphic and apomorphic traits has made the taxonomic placement for these three lineages very uncertain (see Wozencraft, 1984, 1993).
The monotypic genus Nandinia (N. binotata) was placed in the family Nandiniidae by Hunt (1989), McKenna and Bell (1997), and Flynn and Nedbal (1998) and is distributed in the tropical African forests. It was traditionally considered either a peculiar member of the Paradoxurinae, the Asiatic palm civets (Pocock, 1915a), or the representative of the monotypic subfamily Nandiniinae (Gregory and Hellman, 1939; Coetzee, 1977; Wozencraft, 1993). Moreover, Gregory and Hellman (1939) suggested that the trenchant carnassials in Nandinia were "directly derivable from ... [carnassials] ... of the upper Eocene Viverravinae," thus assigning the taxon to a primitive branch antedating frugivorous specialization in Paradoxurinae. Important evidence for plesiomorphic character states and very basal affinities within the Feliformia comes from the structure of basicranium and auditory bulla (Hunt, 1974, 1987). Some external characters also make Nandinia very different from Paradoxurinae and other subfamilies of Viverridae, notably in foot pad and scent gland structures (Pocock, 1915a; Kingdon, 1997). Molecular phylogenetic analyses have placed Nandinia as sister-taxon to all other extant feliformians (Flynn and Nedbal, 1998; Gaubert and Veron, 2003; Yoder et al., 2003).
The three monospecific genera constituting the Malagasy viverrid-like taxa have historically been ascribed to various families of Feliformia because of their heterogeneous morphotypes. The genus Cryptoprocta (C. ferox) has been classified in the Viverridae mainly on the basis of similarities in foot pad structure, palate, and basicrania (Bennett, 1833; Blainville, 1842; Gray, 1864; Flower, 1869; Petter, 1974; Wozencraft, 1989). It has also been placed in the Felidae (Gregory and Hellman, 1939; Beaumont, 1964; Hemmer, 1978) because of its general shape of the skull, hypercarnivorous dentition, and postcranial similarities. In addition, Cryptoprocta has been related to the Herpestidae (Mivart, 1882; Pocock, 1916; Radinsky, 1975; Veron, 1995) according to the presence of an external anal pouch, a long baculum, and a well-developed sulcus cruciatus. The genus Eupleres (E. goudotii) was first described as an insectivore (Doyère, 1835) because of its skull with long rostrum and reduced sharpened teeth (Wozencraft, 1984). It was, however, shortly thereafter included in the feliformians, but assigned to three different lineages: (1) Herpestidae, with which it shares a small paraoccipital process and lack of scent glands (Carlsson, 1911; Petter, 1974); (2) Viverridae, for similar foot pad morphology (Albignac, 1973); and (3) Hemigalinae (especially Chrotogale), on the basis of feeding adaptations (Gregory and Hellman, 1939). Wozencraft (1984) highlighted that Eupleres exhibited plesiomorphic features resembling configurations found in marsupials (lacrimal bone and incisors) and fossil Viverravidae (molars). The genus Fossa(F. fossana) has a less confused taxonomic history because it exhibits basic structures (skull, dentition, foot pad morphology) similar to the Viverridae (Albignac, 1973) or Viverrinae (Wozencraft, 1984). In the last two decades, the Malagasy viverrid-like taxa have been grouped in the Viverridae in the heterogeneous subfamily Cryptoproctinae (Wozencraft, 1984, 1989) or the latter plus Fossinae (Wozencraft, 1993). Recent DNA-based investigations supported the association "Malagasy viverrid-like taxa—Malagasy mongooses (Galidiinae)" as the sister-group of Herpestidae (Yoder et al., 2003).
The genus Prionodon (Asiatic linsangs) consists of two species (P. pardicolor and P. linsang). They share great phenotypic similarity with the African genus Poiana, commonly called the African linsang. These shared features include the hypercarnivorous dentition, spotted coat pattern (although P. linsang is banded), body proportions, and hair ultrastructure (Gaubert et al., 2002a, 2004b; Gaubert and Veron, 2003). Asiatic and African linsangs were first placed in the tribe Prionodontini (Gray, 1864; Mivart, 1882), but subsequent authors considered Poiana a "primitive" genet, which had acquired a morphotype similar to Prionodon by convergence due to their common way of life in tropical rain forests (Gregory and Hellman, 1939; Simpson, 1945; Rosevear, 1974; Crawford-Cabral, 1981, 1993). In addition, Prionodon shares similarities with both Viverrinae and Felidae (Horsfield, 1824; Veron, 1995; Hunt, 2001) that several authors have considered to be evolutionary convergences (Gray, 1864; Pocock, 1933; Gregory and Hellman, 1939). Its basicranial region resembles that of the Oligocene feliformians Paleoprionodon and Proailurus (Teilhard de Chardin, 1915; Gregory and Hellman, 1939; Veron, 1995; Hunt, 2001). Recent molecular investigations suggested the exclusion of Prionodon from the Viverridae (Gaubert et al., 2004b) and placed it as sister-taxon to the Felidae (Gaubert and Veron, 2003), under the family Prionodontidae.
Evolutionary Issues
The study of morphological evolution of viverrid and viverrid-like morphotypes has never been tackled in the context of an exhaustive, solid, and independent phylogenetic framework. However, the great majority of the phylogenetic relationships based on morphology alone often conflict when compared to molecular phylogenies. This was the case for questions like the relationship between Poiana and Prionodon, and the phylogenetic positions of Cryptoprocta and Nandinia within Feliformia (see Veron, 1995; Gaubert et al., 2002a). The recent production of molecular phylogenies concerning feliformians offers the opportunity of comparing a synthetic molecular tree with resolved lineage relationships to morphological evolution. The aim of our study is to (1) characterize the morphological diversification of viverrid-like carnivores from an independent (molecular) phylogenetic hypothesis; and (2) to identify the convergence patterns that may be responsible for the difficulties in reconstructing relationships between viverrids and viverrid-like taxa. For this purpose, a morphological matrix of 349 characters was built, covering a large panel of morphoanatomical regions, and including all species-level viverrids and viverrid-like taxa following Wozencraft (1993). This large matrix allows us to accurately characterize patterns of homoplasy. On this basis, we propose to (1) identify putative homoplastic partitions in the morphological set; (2) to define the mosaic of shared convergences represented by each viverrid-like taxon; and (3) to assess its influence on the phylogenetic reconstruction process. To our knowledge, this procedure has never been applied to such an exhaustive morphological matrix in the framework of a molecular phylogeny. Complementarily, the widespread a priori hypothesis that considers that complex structures either are not subject to or at least less prone to convergence will be tested for four morphological units that have been the subject of intense speculations in feliformian evolution.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
Morphological Data
Morphological data were observed for at least five specimens of each sex (the final number of individuals examined for each species varied depending upon availability of material). We considered juvenile and young adult specimens in order to take into account peculiar characters that were observable in such developmental phases only (see Appendix 1) and to increase accuracy of observation of cranial sutures. Lists of the material examined are available in Wozencraft (1984), Veron (1994), (1995), Gaubert et al. (2002b), and Gaubert (2003a), (2003b).
We coded a morphological matrix consisting of 349 characters partitioned in four subsets: skull (147 characters), dentition (74), postcranium (57), and external/soft anatomy (71) (Appendices 1 and 2, available at the TreeBase website under accession numbers "S1255" [Study] and "M2193" [Matrix]; Appendix 2 also available at www.systematicbiology.org). All characters were described qualitatively (i.e., discrete characters). We coded polymorphic characters when a similar proportion of two different states were observed within a same species. In an attempt to document autapomorphies, all variable characters that appeared to us between species were taken into account. In order to consider the broadest picture of morphological disparity within the Viverridae and viverrid-like taxa, the 34 species recognized by Wozencraft (1993) were included in the analysis (for convenience, we will consider Wozencraft (1993) as the "traditional" classification to which the results of our analyses will be compared and discussed). However, in one deviation from Wozencraft (1993), we placed the aquatic genet in the genus Genetta, following Gaubert et al. (2004a), (2004b). Representatives of the other feliformian lineages were included in the ingroup: Herpestidae (Herpestes, Mungos), Galidiinae (Galidia), Hyaenidae (Hyaena, Proteles), and Felidae (Felis, Lynx). Given the evidences for considering the monophyly of Feliformia (Hunt, 1974; Wyss and Flynn, 1993; Wayne et al., 1989; Wozencraft, 1989; Flynn and Nedbal, 1998) and in order to minimize issues related to character homology, the tree was rooted by three representatives of the suborder Caniformia (Canis, Bassariscus, Ursus), which is the sister-group of the Feliformia. As a consequence of the broad phylogenetic level investigated and our peculiar focus on viverrid-like taxa, the coding of characters could not describe the specific variability prevailing within homogeneous genera such as Genetta, Viverra, and Paradoxurus (studied in Veron, 1995; Gaubert et al., 2002a; Gaubert, 2003a; G. Veron, in preparation).
Phylogenetic Methods
The morphological matrix was first analyzed without topological constraint under PAUP* Beta version 4.0b2 (Swofford, 2001) using the maximum parsimony criteria (MP), with multistate characters treated as "polymorphic." Heuristic searches were performed using stepwise-addition branch-swapping (10 random additions) and tree bisection-reconnection (TBR) swapping algorithm. Phylogenetic signal value (g1; 1,000,000 random trees) was used as an estimation in the structuring of the signal versus random noise (Hillis, 1991; Hillis and Huelsenbeck, 1992) for each matrix subset and in the complete data set. Retention index values (RI; Farris, 1989a, 1989b) were estimated as recommended by Grant and Kluge (2003) as a heuristic method of a posteriori estimation of phylogenetic signal pattern. In order to detect conflicting signals at nodes of interest (i.e., above the species level) between the four partitions, we used the partitioned Bremer support (PBS; Baker and DeSalle, 1997) with TreeRot.v2 (Sorenson, 1999). We then evaluated the significance of each branch support value by employing the test of Templeton (Macey et al., 1999; Whitlock and Baum, 1999; Lee, 2000; Lee and Hugall, 2003).
A topological constraint was then imposed on the phylogenetic search in order to trace back morphological evolution from an independent topology and to detect putative morphological convergence. This constraint corresponded to a synthetic tree built by hand from recent works on the evolution of feliformians and including viverrids (Flynn and Nedbal, 1998; Veron and Heard, 2000; Gaubert and Veron, 2003; Yoder et al., 2003; Gaubert et al., 2004a, 2004b). Only nodes with strong bootstrap (> 75%) or posterior probability values (> 0.95) and found throughout the different studies were considered (Fig. 1). Several viverrid species were not represented in molecular studies (especially members of the Paradoxurinae and Hemigalinae), and so their relationships were depicted as polytomies in the constraint tree. Sufficient taxonomic sampling and phylogenetic support were, however, available to fix dichotomous, unambiguous branching patterns between all lineages, reaching the subfamily level in Viverridae. More than half of the nodes of concern (A, C, D, F, G, H, and K; see Fig. 1) were supported by published analyses based on a total number of four independent loci (mtDNA: cytohrome b and ND2/nDNA: transthyretin intron I and IRBP). In what concerns the Viverridae, nodes H, J, K, L, and M were also found with high support using a second nuclear locus (totalizing 1 mtDNA and 2 nDNA genes; Gaubert et al., unpublished data). Thus, we made the assumption that both the nature and the high level of homoplasy in the morphological data set (Wyss and Flynn, 1993; Veron, 1995; Flynn and Nedbal, 1998; Veron and Heard, 2000; Gaubert and Veron, 2003; Yoder et al., 2003; Gaubert et al., 2004b) were more likely to yield a misleading phylogeny when compared to the congruent studies underlying the molecular synthetic tree.
|
Characterization of Convergence and Phylogenetic Perturbation
As a first attempt at detecting homoplasy, we determined the fluctuations of RI values between the constrained (RIc) and unconstrained (RIu) topologies. The characterization of RI distributions (i.e., values of RI for a set of characters classified from 0.0 to 1.0) were determined for the complete matrix and each separate subset in both tree search conditions using a density estimation function of the stats library of the R statistical language (Ihaka and Gentleman, 1996) version 1.9 for Linux (http://cran.r-project.org/). The purpose of density estimation is to model the distribution of RI values on the basis of sample distributions. Then, we assessed the perturbation due to the imposition of a topological constraint by statistically testing the differences of behavior between RIc and RIu distributions. Because RIc and RIu cannot be considered as independent values, we applied the paired-sample hypothesis testing case. We first tested the normality of the distribution of the (RIc – RIu) vector through a Shapiro-Wilk test (Royston, 1982a, 1982b, 1995), and a paired t-test was used each time the different partitions met this assumption (Zar, 1984). When the hypothesis of normality in distribution was rejected, we used the nonparametric Wilcoxon paired sample rank test (Bauer, 1972; Hollander and Wolfe, 1973).
The characterization of homoplasy "caused" by the viverrid-like taxa was calculated as the number of shared convergences per taxon for each partition of the constrained analysis. We used the "Describe trees—list of apomorphies" option in PAUP*, and calculated the number of shared convergences between each viverrid-like taxon and the following list of lineages, (in which they have been historically included): Felidae, Viverridae, Paradoxurinae, Viverrinae, Hemigalinae, Herpestidae. Similar comparisons were also made between each genus of viverrid-like taxa. Because the Malagasy viverrid-like taxa grouped as (Cryptoprocta, (Eupleres,Fossa)) in the constrained analysis, we also considered the number of shared convergences at the branches leading to the two nodes whenever these acquisitions were shared by all the taxa of a node (the same procedure was applied at the node grouping the Felidae and Prionodontidae). Given that different optimizations did not affect the number and nature of shared convergences, we present here the results from ACCTRAN optimization. The mosaic of convergences constituted by the viverrid-like taxa was illustrated as a percentage representation of the sum of convergent character states shared with the main lineages of feliformians (including Viverridae subfamilies). Diagrams based on raw calculations of shared convergences were used in order to give a detailed representation of the contribution of each morphological partition to the mosaic.
In order to estimate the level of perturbation caused by viverrid-like taxa in the phylogenetic reconstruction, we re-coded the identified convergent states as uncertainties (i.e., coded "?") in each of the five taxa, thus allowing "free" optimization during tree search. We first ran the analysis with one modified taxon, the other viverrid-like taxa remaining unchanged. Then, two, three, and four taxa had their coding alternatively modified, until the five taxa were all included as modified in the matrix (25 alternative matrices). In each case, lengths were compared among the constraint tree and (1) the unconstrained trees and (2) the trees based on modified coding using the Templeton (Wilcoxon signed-ranks) test (Templeton, 1983), as implemented in PAUP* and following Graham et al. (1998) when several most-parsimonious trees were available. This is a nonparametric, conservative test, which relies on the calculation of differences in number of steps involved by two competing topologies for each character (Cunningham, 1997). We assumed that a significant departure from initial tree length (unconstrained analysis) was evidence for hidden phylogenetic signal, the contribution of which could have been hampered by convergences. On the other hand, nonsignificant length difference with the constrained tree was assumed to correspond to the identification of mosaics of convergence misleading the search of the "true" tree topology.
Test of Evolutionary Hypothesis
The mapping of morphological units onto the strict consensus tree obtained from the molecular constraint was done using linear parsimony under MacClade 4.0 (Maddison and Maddison, 2000). This method uses Wagner parsimony (Swofford and Maddison, 1987) and opts for the scenario that minimizes the number of evolutionary events necessary to explain the distribution of character states on the tree (Maddison and Maddison, 1992; Cunningham et al., 1998). In order to give a relative support to the scenario, we assigned the alternative character state that implied the least parsimonious option at the base of the feliformian clade, and calculated the number of additional evolutionary events on the tree. We then estimated the ratio between the latter value and the number of steps for the ancestral condition that implied the most parsimonious option. We thus obtained an index that reflects the proportional difference between the most and the least parsimonious choice of ancestral character states (Gaubert et al., 2004a). We chose to test the evolution of four complex functional units (Appendix 3, available at www.systematicbiology.org) in order to reassess their evolutionary history using an independent and robust phylogeny. Complex structures have been considered as "reliable" phylogenetic characters because the great amount of coordinated transformations that is necessary to evolve is supposed to minimize convergences (e.g., auditory bulla in feliformians: Hunt, 1974, 1987, 1991; Hunt and Tedford, 1993). Various hypotheses have been proposed concerning the orientation of transformations in complex functional units and what conditions are primitive in feliformians, especially viverrids. We thus compared traditional assertions to the recent clarification of feliformian phylogenetics by specifically retracing the evolution of four complex structures:
- Foot structure (plantigrade–semi-digitigrade–digitigrade). We follow the definition of Taylor (1988) in considering that plantigrady is characterized by relatively short metapodials with distinct metapodial pads whereas relatively long metapodials with reduced or vestigial metapodial pads are diagnostic of digitigrady. Semi-digitigrady is specific of the genera Genetta and Poiana, which exhibit plantigrade forefeet and digitigrade hindfeet. Data collected from literature (Pocock, 1915a, 1915b, 1915c, 1915d, 1915e, 1916a, 1916b, 1916c, 1916d; Ginsburg, 1961; Taylor, 1988) were compared, for all species, to direct observations on museum specimens (Wozencraft, 1984; Veron, 1999; Gaubert et al., 2002a).
- Perineal glands: (absent–present [Nandinia type]–present [viverrine type]–present [paradoxurine type]–present [hemigaline type]). Given the relative heterogeneity and scarcity of bibliographical sources on the structural types of perineal glands in feliformians, we coded the different types following two conservative criteria: (1) relative position of the glands to the urogenital orifice and (2) inflation of the glandular fossa. The first type corresponds to the peculiar condition met in Nandinia, with the urogenital orifice at the posterior part of the glandular area, the anterior part of which forms a shallow subcircular depression with tumid margins. The secreting cells are not differentiated into a right and left thickening like in other taxa where perineal glands are present. The second kind defines the Viverrinae, with the glands occupying a strictly perineal position and divided in two distinct pouches for the storage of secretion. The third type is typical of the Paradoxurinae, which have the urogenital orifice encircled by the labia of the glands, either in central or anterior position. The perineal glands form two distinct pouches as in the Viverrinae, but more flattened. Finally, the fourth kind corresponds to the state met in at least some of the Hemigalinae (data available for Hemigalus and Cynogale), with the urogenital orifice encircled by the most anterior part of the labia of the glands, the latter showing "shallow glandular pits" (Pocock, 1915b, 1915d) in which secretion is stored. Although perineal glands are clearly present in Galidia (Pocock, 1915f; Nowak, 1999; WCW personal observation), the bibliographical sources were too confusing to envisage any accurate characterization and Galidia was coded "?" (see Discussion). Data were collected from the literature (Pocock, 1915a, 1915b, 1915c, 1915d, 1915e; Ewer, 1973) and, depending on the availability of the material, from personal observations (see Wozencraft, 1984; Gaubert, 2003b).
- External anal pouch (absent–present [central anal sac]–present [anterior anal sac]). State "absent" corresponds to functional anal glands, but not resulting in external inflation. The central anal sac is the condition met in Cryptoprocta and most herpestids, which exhibit large anal glands opening in a pocket-like anal sac surrounding the anus. The anterior anal sac characterizes the Hyaenidae, where the condition is similar but the anal sac is situated anterior to the anus. Data were exclusively collected from the literature (Pocock, 1916a, 1916b, 1916c, 1916d; Ewer, 1973; Gorman et al., 1974; Wyss and Flynn, 1993).
- Tympanic bulla conformation (types "A" through "G"). We follow Hunt (1974), (1987), (1991), (1998), (2001) and Hunt and Tedford (1993) in defining the different types of bulla conformation on the basis of the relative proportions of the two bulla chambers (i.e., ectotympanic and caudal entotympanic bones). Type A is the condition met in several lineages of arctoids (including Ursus and Bassariscus), with a very small, bipartite entotympanic relative to ectotympanic, and no true septum. Type B is found in canids and other arctoids, and exhibits a single, enlarged caudal entotympanic with occasional pseudoseptae that partially subdivide the bulla. Type C is exclusive to Nandinia, with a cartilaginous caudal entotympanic and only slightly inflated. The septum is absent. Type D characterizes the Viverridae and viverrid-like taxa, with an enlarged caudal entotympanic expanded forward into the anterointernal corner of the auditory region, and a vertical septum bulla. Type E is the condition met in Hyaenidae, where the caudal entotympanic is reduced and the ectotympanic very large. The bulla septum is recumbent (Hyaena) or vertical (Proteles). Type F is found in Felidae. It is similar to Type D (Viverridae) with the exception of the caudal entotympanic, which grows forward during early stages of development to cover the ectotympanic–rostral entotympanic contact. Type G is typical of Herpestidae and Galidiinae. It shows a generally vertical septum bulla, and the anterior chamber of caudal entotympanic is aligned in front of the posterior chamber of ectotympanic with no encroachment (excepted in Cynictis). Whenever possible, data collected from literature were cross-checked by direct observations on museum specimens.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
Phylogenetic Analysis without Topological Constraint
The exhaustiveness of our morphological observations yielded a low percentage of missing data (< 5%) in the final matrix, with six exceptions: Macrogalidia musschenbroekii (7.7%), Genetta abyssinica (12.3%), Paradoxurus jerdoni (13.2%), Diplogale hosei (14.6%), Chrotogale owstoni (14.9%), and Poiana richardsonii (23.8%).
The phylogenetic analysis of the morphological matrix yielded eight equally parsimonious trees (1761 steps; CI = 0.331; RI = 0.635), resulting in a strict consensus tree of 1782 steps (Fig. 2; CI = 0.327; RI = 0.628). Estimations of phylogenetic signal values (g1) by partition ranged from –0.388389 (postcranium) and –0.360938 (external/soft anatomy) to –0.450502 (skull) and –0.454838 (dentition) (g1total = –0.439344). This reflected weak skewness in tree-length distributions (critical values not available for morphological data sets including nonbinary characters; see Hillis and Huelsenbeck, 1992). The topologies of the consensus and synthetic molecular trees were highly dissimilar. None of the phylogenetic relationships established by recent molecular studies were recovered between the main feliformian lineages. Node support values were generally very low, but reached high values in a few exceptions such as the monophyly of Hyaenidae, Felidae, and Asiatic linsangs (Fig. 2). The Hyaenidae and Felidae formed a sister-group clade to the other feliformians. The ingroup then split into (Herpestidae, Galidiinae) and a clade that was historically grouped under "viverrids" (i.e., Wozencraft, 1993). The latter consisted of three distinct lineages. The first was the two Malagasy viverrid-like taxa (Eupleres and Fossa). The second consisted of Cryptoprocta(Malagasy viverrid-like taxa) followed by Nandinia (another viverrid-like), then the Paradoxurinae (here paraphyletic) and Hemigalinae (monophyletic) as the crown group. Finally, the Viverrinae as they were traditionally defined formed the third lineage. The viverrid clade was supported by the only diagnostic synapomorphy of the analysis, namely the presence of three facets on the head of the malleus (ossicles; 126(1)).
|
The respective contribution patterns of each character partition proved to be heterogeneously distributed throughout the phylogenetic tree. The "skull" partition significantly supported the monophyly of the viverrids whereas this subset did not support the majority of the nodes within the group. However, two notable exceptions included the monophyly of Prionodon and Civettictis + Viverra. The phylogenetic signal conveyed by dental characters supported several nodes within the Paradoxurinae, Hemigalinae, and terrestrial civets. Postcranial anatomy supported the monophyly of the viverrids, as well as the association (Cryptoprocta, (Nandinia, (Paradoxurinae + Hemigalinae))) and ((Poiana, Prionodon), Genetta). The partition "external/soft anatomy" was the most important in terms of tree structuring. It significantly contributed to the monophyly of Poiana + Prionodon and was a strong support for the traditional grouping of the Viverrinae. It contributed to the greatest support of the monophyly of the terrestrial civets and Hemigalinae. The external/soft anatomy also supported the branchings of Cryptoprocta and Nandinia at the base of their clade.
Phylogenetic Analysis with Topological Constraint
The constrained analysis yielded two most-parsimonious trees (1869 steps; CI = 0.312; RI = 0.601) resulting in a strict consensus tree of 1871 steps (CI = 0.312; RI = 0.600). Morphological characters supported similar relationships in the Hemigalinae and terrestrial civets compared to the unconstrained analysis. The Paradoxurinae grouped this time as ((Macrogalidia, (((Paguma, (Arctictis, Arctogalidia)), Paradoxurus)), and the Malagasy taxa grouped as (Galidia, (Cryptoprocta, (Fossa, Eupleres))).
The topology of the unconstrained tree was significantly more parsimonious than the constrained tree (Templeton's test: P < 0.0001). The increase in number of steps induced by the topological constraint was significant for each character partition (P < 0.05). When the unconstrained tree topology was compared to the tree built via matrix representation using parsimony of Bininda-Emonds et al. (1999), length differences based on the complete data matrix were significant (MRP tree constraint: 1802 steps; P = 0.0006). However, the Templeton's test, based on the "postcranium" and "external/soft anatomy" partitions, did not detect significant length differences. A similar incongruence between the phylogenetic relationships of feliformians proposed by Bininda-Emonds et al. (1999) and the topology of the molecular (constraint) tree also was found (P = 0.0001). In this case, the subsets "dentition" and "postcranium" did not involve significant length differences.
In contrast to the low number of diagnostic synapomorphies characterizing the unconstrained tree, the constrained analysis resulted in the identification of numerous, exclusive homologies, including the Feliformia, Herpestidae, Hyaenidae, and Felidae (Table 1). A new synapomorphy for the recently erected family Prionodontidae (Gaubert and Veron, 2003) was identified in the fusion into one straight "keel" of the spinous processes of the sacrum vertebrae (postcranium: 231(1)). The sister-group relationship between the Paradoxurinae and Hemigalinae was supported by a single synapomorphy, which consisted of the symmetry of M1 buccal lobes (dentition: 185(1)). The presence of a hypocone on M2 was diagnostic for the monophyly of the Hemigalinae (dentition: 197(1)). The clade (Civettictis, Viverra) was characterized by a diagnostic synapomorphy from postcranial characters, namely the convex profile of the axis spinous process (227(1)). On the other hand, no diagnostic morphological synapomorphies were found for all the interlineage relationships recently proposed on the basis of molecular studies.
|
Characterization of Convergence and Phylogenetic Perturbation
Density estimation analysis revealed a trimodal distribution of RIs in both tree search conditions for the whole set of morphological characters, with three peak values at 0.0, ca. 0.7 and 1.0 (Fig. 3, "Complete matrix"). The Wilcoxon paired sample rank test detected a highly significant decrease in RIs from the unconstrained to constrained analyses (P < 0.0001). RI distribution patterns were different among the four character partitions both in the unconstrained and constrained analyses, but all exhibited a main peak between 0.5 and 0.8. The postcranium partition was the only one to match the normality criterion. "Skull" and "external/soft anatomy" subsets showed a bimodal distribution, whereas "dentition" was unimodal. The "potscranium" partition had a different behavior depending on tree search conditions, because RI distribution, shifted from unimodal (unconstrained) to trimodal (constrained). The four partitions revealed a significant decrease in RIs (P < 0.0001; postcranium: P < 0.00197) when the topological constraint was enforced, but the difference RIc – RIu generally did not exceed –0.2. There was no correlation between initial RI values derived from the unconstrained search and their posterior variability (data not shown). The bimodal patterns characterizing the "skull" and "external/soft anatomy" partitions corresponded to the first two modes described by the complete data matrix: a small peak at 0.0, a decrease in RI frequencies between 0.1 and 0.3, and a main peak between 0.4 and 0.8. On the contrary, the "dentition" subset did not show evident multimodal structure. Relevant changes observed in the distribution of RIs when the analysis was constrained appeared in (1) the "skull" partition, with a decrease in the frequency of RIs concerning the least homoplastic characters; (2) the "dentition" subset, which showed an increase in characters with RI ranging from ca. 0.4 to 0.5; and (3) the "postcranium," which was characterized by a shift from unimodal to trimodal structure (peaks: 0.0, 0.6, 1.0) and an increase at the main peak (0.6).
|
The proportion of characters that were more homoplastic after the constrained search (RIc – RIu < 0) was higher in the "postcranium" (38.6%) and "external/soft anatomy" (42.3%) partitions (Fig. 4) than in "skull" and "dentition" (< 30%). On the contrary, the proportion of characters with less homoplasy was roughly constant in each partition (11.6% to 17.5%). In all cases, the part of characters with neutral behavior (RIc – RIu = 0) was the most important (42.3 to 60.3%). Note that autapomorphies (RIc – RIu always equals 0) were included in the calculations and did not affect the results (9 [skull], 8 [dentition], 3 [external/soft anatomy]), excepted in the case of the external/soft anatomy partition where the "real" neutral set represent 28 characters (thus slightly lower than characters [30] with RIc – RIu < 0). Dramatic changes in level of homoplasy (RIc – RIu = –1) affected characters 47 (postorbital inflation of frontal), 286 (shape of external pinnae), and 349 (presence of marker chromosome).
|
The calculation of proportions of shared convergences for each viverrid-like taxon yielded a pattern in mosaic, which was remarkable by the fact that a great part of homoplasies were shared among viverrid-like taxa themselves (Fig. 5; Appendix 4, available at www.systematicbiology.org). The genus Nandinia shared the greatest proportion of convergences (n = 77) with Cryptoprocta (19%), Fossa, and Prionodon(both at 16%). Nandinia also shared 13% of homoplasies with Eupleres and the Paradoxurinae. The global pattern of convergence observed in Cryptoprocta (n = 54) was principally explained by the contributions of Nandinia(28%), the Felidae (22%), Paradoxurinae (17%), and Prionodon (15%). Cryptoprocta shared only 2% (n = 1) of its homoplastic characters with the Herpestidae. The genus Fossa had a majority of convergent states (n = 45) shared with Nandinia (26%), the Viverridae (18%), Prionodon, and the Viverrinae (both 16%). The mosaic pattern in Eupleres (n = 46) was mainly explained by shared convergences with Prionodon (32%) and, to a lesser extent, Nandinia (22%). The genus Prionodon had a mosaic of shared homoplasies (n = 72, including Poiana; see Fig. 5) to which Eupleres (20%), Nandinia (17%), and Poiana(13 %) mainly contributed. We then assessed the respective contribution of each character partition to the patterns of shared convergences observed among viverrid-like taxa. This revealed a second level of mosaic, which was distributed in this case among the four data subsets (Fig. 6). In Nandinia, the number of convergences related to skull was almost equally shared (6–7) with the other viverrid-like taxa, whereas Cryptoprocta supplied the highest number of shared convergences in postcranium (7). The most significant contributions found in Cryptoprocta were shared convergences with the Felidae in dentition (9), and Nandinia in skull (5), and postcranium (7). The genus Fossa shared convergences related to the skull partition supplied by Nandinia (6) and Prionodon (5). The Viverridae contributed to most of the convergent character states in dentition (3), and Nandinia also supplied the highest contribution for external/soft anatomy (4). In Eupleres,Prionodon principally contributed to skull (11), whereas convergences in the other partitions were diffusely distributed. The most significant values for Prionodon were provided by Eupleres (skull; 11), Nandinia(postcranium; 5), and Poiana(dentition; 3).
|
|
The different levels of perturbation applied to shared homoplasies between viverrid-like taxa yielded tree lengths (1) not significantly different from the initial, unconstrained phylogeny (P > 0.05 in all cases) and (2) greatly different from the constrained phylogeny (P < 0.0001 in all cases; data not shown). None of the obtained topologies recovered viverrid-like taxa branchings as found in the synthetic molecular tree.
Test of Evolutionary Scenarios for Complex Structures
Based on the mapping of traits related to the foot structure, the constrained phylogenetic tree suggested that the plantigrade condition was ancestral to the feliformians (Fig. 7; index values [least/most parsimonious scenarios] for semi-digitigrady and digitigrady = 7/5 and 6/5, respectively). Three reversions appeared (digitigrade 
plantigrade), namely in Cryptoprocta, the Hemigalinae, and between the Paradoxurinae node and the branch leading to the common ancestor of the Paradoxurinae and Hemigalinae (ambiguous reconstruction). Semi-digitigrady strictly characterized the clade Poiana + Genetta (Viverrinae). The absence of perineal glands was considered the ancestral state in feliformians ([Nandinia type] = 6/5; [viverrine, paradoxurine and hemigaline types] = 7/5). The pattern of acquisition for perineal glands characterizing the Viverrinae, Paradoxurinae, and Hemigalinae was obscured by ambiguous reconstruction in the branches leading to their common ancestors. The uncertainty related to the coding of Galidia (see Fig. 7) did not allow us to distinguish between acquisition of perineal glands as a homoplastic event or autapomorphy. Similarly, we could not assess whether the absence of perineal glands in Poiana, ascertained for the first time in this study, is a symplesiomorphy shared with Prionodon and other fielformian lineages or a secondary loss. The absence of an external anal pouch was estimated as the ancestral condition in feliformians (Fig. 8; [central and anterior anal sacs] = 6/3). This structure was subject to one reversion in (Fossa, Eupleres) ([central anal sac] absent), but the condition in Mungotictis is unknown. The reconstruction of the ancestral state concerning tympanic bulla conformation in the feliformians was ambiguous (Fig. 8), with types A, B, C, and D having equal probabilities (types E, F, and G = 8/7). Type D was estimated as the symplesiomorphic state characterizing the feliformians excluding Nandinia. Type F (Felidae) appeared derived from type D, whereas ambiguous reconstruction on the branch leading to the common ancestor of the Hyaenidae, Herpestidae, and Malagasy taxa hampered clear reconstruction (at least three new acquisitions can be envisaged).
|
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
Characterization of Convergence: Viverrid-Like Taxa as Non-Adaptive Morphological Mosaics
We assumed that well-supported clades found iteratively in molecular analyses (notably from independent regions of the mitochondrial and nuclear genomes; see Fig. 1) constituted more consistent estimates of phylogenetic relationships when confronted to the results from the morphological analysis. This would be equivalent to the "third explicit criterion" proposed by Wiens et al. (2003) for detecting misleading morphological convergence. The unconstrained analysis based on the morphological matrix yielded phylogenetic groupings that remarkably reflected the traditional classification in use during the past decades in grouping all the viverrid-like taxa in the viverrids (Wozencraft, 1984, 1989; Corbett and Hill, 1991). The historical proximity of Nandinia to the Paradoxurinae was reflected by its position at the base of the clade (Paradoxurinae-Hemigalinae). This node was mainly supported by the external/soft anatomy partition, although our coding included the particular traits of Nandinia in pad and scent gland structures as noticed by Pocock (1915) and Kingdon (1977), (1997). The three Malagasy viverrid-like taxa were not grouped together in a single taxon (Wozencraft, 1984, 1989). However, the genera Eupleres and Fossa constituted a unique lineage, whereas Cryptoprocta, on the basis of the dentition and external/soft anatomy partitions, was sister-group of (Nandinia, Paradoxurinae-Hemigalinae). The clade (Prionodon, Poiana) was significantly supported by external characters whereas the skull partition was in conflict. The phylogenetic relationships obtained from our data matrix and the MRP (supertree) analysis of Bininda-Emonds et al. (1999) both reflect, at different levels, the traditional taxonomy of "viverrids." They are typical illustrations of morphological phylogenies driven by an "all combined" approach of data sets (i.e., no a priori removal of potentially convergent data; see McCracken et al., 1999; Wiens et al., 2003). The supertree of the Carnivora proposed by Bininda-Emonds et al. (1999) has already been used as a reference topology for testing macroevolutionary patterns and trends (e.g., Sechrest et al., 2002). Following Ruta et al. (2003), we urge caution in (1) using supertrees that include all source trees available (i.e., even ones that have been superseded by more comprehensive studies), and (2) using supertrees as support for replacing character-based phylogenies.
The use of a synthetic molecular tree as a constraint allowed us to analyze morphological evolution from independently derived phylogenetic hypotheses. One of the striking points of our results resides in the mosaics of convergences that the viverrid-like taxa constitute: these convergences being mainly shared among viverrid-like taxa themselves. The complexity of this homoplastic pattern was increased by the existence of a second level of mosaic concerning the contributions per taxa of the four data partitions. Thus, each viverrid-like taxon proved to represent a complex and coherent combination of convergences. These results demonstrate that considering "rough" phenotypic similarities between morphotypes (i.e., "coat pattern," "dentition," etc.) may lead to misleading assumptions regarding their respective contributions to convergence. For instance, the genus Prionodon, often mentioned as representing an extreme case of convergent morphotype with Poiana, shares instead an unsuspected, high proportion of convergences with the other viverrid-like taxa. The hypothesis put forward by Gaubert and Veron (2003) about global convergence between the skull morphology of Poiana and Prionodon needs to be reconsidered in the light of the exhaustiveness of our analysis; given that the "skull" partition did not support the association between Asiatic and African linsangs, and most of the cranial convergences are shared with Eupleres. Our results illustrate that long-standing debates over the classification of viverrid-like taxa within the Feliformia were at least partly generated by the unsuspected mosaics of convergences they share between themselves and other feliformian lineages, in addition to symplesiomorphies (e.g., Ginsburg, 1961; Petter, 1974; Hunt, 2001). We consider that the unique mosaics of convergences characterizing viverrid-like taxa constitute further evidence for proposing their classification into three distinct families: Nandiniidae (Nandinia), Prionodontidae (Prionodon), and, as a newly defined family, Eupleridae Chenu, 1852 (Cryptoprocta, Fossa, Eupleres + "mongoose-like" Malagasy taxa: Galidia, Galidictis, Mungotictis, Salanoia). The inclusion of all the Malagasy taxa in a single family is justified by their well-supported monophyly (Yoder et al., 2003) and a drastic morphological divergence compared to their sister-group, the Herpestidae. However, their significant morphological heterogeneity, once relationships among viverrid-like Malagasy lineages are clarified, might justify a split into several family level taxa.
Morphological- and molecular-based phylogenies yielded dramatically incongruent topologies. However, no clear "phylogenetically misleading" morphological data could be identified from the comparison between the constrained and unconstrained tree searches. Our analysis showed that the respective contributions of the four different data sets along the unconstrained tree were distributed heterogeneously. For instance, the "skull" partition significantly supported grouping Viverridae and viverrid-like taxa in a single clade and supplied one diagnostic synapomorphy at this node; whereas external characters had an important contribution at nodes designing traditional groups of lower taxonomic levels (Viverrinae, Poiana + Prionodon). Similarly, in the constrained analysis, the contribution of character subsets proved dependant on the taxonomic level of investigation (as suggested for dentition by Hunt and Tedford, 1993). This was illustrated by the diagnostic synapomorphies supplied by the "skull" partition at the family and suprafamily levels (Feliformia, Felidae, Hyaenidae); whereas the "dentition" subset principally contributed to infrafamilial structuring (Hemigalinae, Paradoxurinae + Hemigalinae).
In the case of incongruence between morphological and molecular phylogenies, adaptive morphological data subsets involving convergence are often attributed the role of misleading characters (e.g., Alvarez et al., 1999; Littlewood et al., 1999; McCracken et al., 1999; Goodacre and Wade, 2001; Wiens et al., 2003). Conversely, our analysis suggests that the great majority of morphological convergences are not adaptive. If we focus on the most homoplastic traits (i.e., convergences shared more than once among viverrid-like taxa; Table 2), almost none of them can be associated with what would appear to us to be adaptive functions. For instance, although the independent acquisition of metatarsal pads larger than tarsal pads (311(3)) in Nandinia, Cryptoprocta, and Paradoxurinae is probably indicative of adaptation to arboreality, how can one interpret the shared convergence of the absence of a hypoconid in M1 (192(1)) between Nandinia(omnivorous) and Cryptoprocta and the Felidae (both hypercarnivorous)? A possible explanation to this apparent lack of adaptive value in morphological characters may reside in the fact that there are several ways of adapting to a peculiar mode of life, diet or locomotion (Leal et al., 2002), including the use of identical traits for different functions (Table 2). A second potential explanation is that the peculiar patterns in shared convergences observed in viverrid-like taxa are due to "phylogenetic inertia," i.e., nonadaptive phenotypic stasis acting heterogeneously along the tree (Wilson, 1975). A developmental perspective to this hypothesis is that independent acquisition of presumably nonadaptive characters may be due to recurrent processes of canalization (Zakharov, 1992; Debat and David, 2001). In addition, canalization may interact with other iterative developmental processes influencing the morphotype expression. For instance, it has been shown that only small heterotopic shifts and iterative processes in the molecular prepatterns of gene expression were necessary to explain the observed diversity in the molars of mammals (Jernvall et al., 2000). Moreover, similar morphologies can be produced by changes targeted on different regions of the gene network affecting morphogenesis (Salazar-Ciudad and Jernvall, 2002). A third alternative is that we simply do not know enough of the implications of each character in the biological functions of the studied organisms. It may be probable that our failure to identify any adaptive trends in the mosaics of shared convergences result in a mixture of these three potential explanations.
|
Noise Pattern in Morphological Data: Why a Misleading Phylogeny from Morphological Characters?
Our results suggest that, rather than convergences, the responsible factor for the misleading tree topology obtained using the morphological matrix is the absence of structuring phylogenetic signal. Several evidences extracted from our empirical study support the hypothesis that noise (i.e., random data distribution; Wenzel and Siddall, 1999) is predominantly present in the data set. Potentially the best way to estimate the presence of phylogenetic signal is to look at the number and distribution of diagnostic synapomorphies along the tree (Wenzel and Siddall 1999; Mattern and McLennan 2000). In our case, only one diagnostic synapomorphy was supplied by the whole data set in the framework of the unconstrained search (node "viverrids") and the Bremer support values were in general very low or not significant. The use of partitioned Bremer supports allows us to discard the hypothesis of low node supports as a result of conflicting phylogenetic signals, because no significantly antagonistic PBS values were found between the four data partitions. Furthermore, the length distributions of randomly distributed tree topologies (g1) for the complete matrix and each partition were not diagnostic of the presence of structuring phylogenetic signal (low g1 values, distribution pattern not clearly left-skewed). This hypothesis is also supported by the low amplitude (although significant) decrease in RI values induced by the constrained search and by the neutral behavior of the major part of the characters to this constraint. Finally, the fact that successive removals of convergences did not induce significant changes towards the constrained phylogeny nor from the initial tree discards the hypothesis of shared homoplasies acting as misleading characters in the unconstrained tree. It has also been suggested that the lack of phylogenetic signal may be due to a rapid diversification of the taxa examined (Mardulyn and Whitfield, 1999). However, multiple evidences from fossil records and molecular date estimates suggest that the main lineages of feliformians, including viverrid-like taxa, diverged during a long period between ca. 40 and 20 Mya (Hunt, 1989, 1991, 1996; Veron and Catzeflis, 1993; Gaubert and Veron, 2003; Yoder et al., 2003). In addition, the fact that our matrix includes a very exhaustive taxonomic sample set tempers the possibility of obtaining noise induced by incomplete taxonomic sampling (Hillis, 1996, 1998; Purvis and Quicke, 1997), although the inclusion of fossil taxa—not done here—may dramatically affect phylogenetic reconstruction and character behavior (Werdelin, 1996). From our analysis, we suggest that low phylogenetic structuring is the result of a combination of stochastic evolutionary processes and of constraints related to phylogenetic inertia/canalization.
Density estimation function analysis of RI distribution could allow a first step towards the characterization of noise. RI values for the whole data set followed a trimodal distribution, which consisted in an assemblage of character subsets with different behaviors. Some dramatic changes in distribution patterns were induced by the constrained search, but the different partitions did not react similarly to topological enforcement. Together with the presence of numerous diagnostic synapomorphies (constrained search), this may suggest that phylogenetic signal, even weak, is present but distributed heterogeneously within the data set. Thus, noise might confuse phylogenetic reconstruction by overwhelming "true" phylogenetic structuring, and the opposition "additive signal versus averaged noise" (Wenzel and Siddall, 1999) seems not that simple. Our results also exemplify the complexity of the task when attempting to distinguish noise from signal in a composite partition (Wenzel and Siddall, 1999). Although it remains hazardous to fix RI-based thresholds for categorizing noise (but see Zaragüeta i Bagils et al., 2002), our approach using density estimation function may be useful in identifying groups of "noisy" characters (e.g., these could be removed from the analysis each time the same character sets in both unconstrained and constrained analyses peak at 0.0).
The apparent stochasticity of morphological evolution may partly or totally arise from the subjective coding of discrete characters, hardly avoidable in morphological analysis (Stevens, 1991; Smith, 1994; Graham et al., 1998; Scotland et al., 2003). For instance, our data matrix consists of almost 50% (171) of nonbinary characters, thus implying "discretization" made by the observer which may rend search for phylogenetic signal and hierarchic structure illusive (Pleijel, 1995; Wilkinson, 1995; Forey and Kitching, 2000; Hawkins, 2000). In order to test for the putative misleading influence of nonbinary characters in the tree search, we ran an analysis based on a matrix restricted to the 178 binary characters. We obtained 244,930 most-parsimonious trees (552 steps; CI = 0.322; RI = 0.658), resulting in a strict consensus tree of 566 steps (CI = 0.314; RI = 0.645) that was congruent with the tree topology issued from the complete matrix, and had no significantly supported nodes other than those found in the initial analysis (data not shown). Thus, although nonbinary characters may constitute an open door to additional noise, they could not be incriminated in our case. Biased estimations of phylogenetic signal may also result from our authoritative partitioning into four data subsets, the subjectivity of partition delimitations in morphology also remaining hardly avoidable (Kluge, 1989, 1997; but see McCracken et al., 1999). We tried to follow the partitioning of Wyss and Flynn (1993) and to maintain a comparable number of characters in each partition, thus avoiding possibly misleading splitting into small subcategories. Besides, we envisaged each partition as a unit of complex interacting entities. For instance, although segregating between bulla and the other cranial bones would have been tempting, the acknowledged interaction, notably during early phases of development, between bones constituting the bulla and adjacent bones of the basicranium (Hunt, 1974, 1987; Ivanoff, 2001; Peigné and De Bonis, 2003) made us opt for a conservative grouping under the single "skull" partition. In the future, the consideration of genomic-based developmental criteria for morphological character delimitations should be a crucial tool for improving the accuracy of data coding and partitioning in phylogenetic analysis (e.g., Salazar-Ciudad and Jernvall, 2002; Hlusko, 2004).
Evolution of Complex Structures
Mapping characters on the constrained tree topology and the reconstruction of ancestral conditions using parsimony clearly suggested that complex morphological structures may not be evolutionary "reliable" (i.e., not homoplastic) entities. The ancestral trait in foot structure at the base of feliformians was infered to be plantigrady, which is in agreement with the fossil record concerning feliformians from Late Eocene–Early Oligocene (Ginsburg, 1961; Ewer, 1973). Our analysis showed that three reversions appeared in foot structure condition, which is a strong argument against the phylogenetic stability of this functional unit (Ewer, 1973; Taylor, 1988). Alternatively, it would be tempting to attribute an ecological association between foot structure and mode of life (Ewer, 1973; Taylor, 1976). For instance, the Paradoxurinae, Nandinia, and Cryptoprocta are plantigrade and arboreal, whereas Fossa, Eupleres, and terrestrial civets are digitigrade and terrestrial. However, some challenging cases remain, as exemplified by the arboreal taxa Prionodon (digitigrade) and Poiana (semi-digitigrade). Adaptation to arboreality may represent secondary events in the history of these taxa, subsequent to the acquisition of plantar structures. In view of these results, the causality of the several reversions that occurred among the Viverridae and viverrid-like taxa remains unclear. At least, locomotor diversity found in Miocene fossil felids confirms the lability of foot structure (Ginsburg, 1999).
The evolution of the external anal pouch also proved homoplastic, with Fossa and Eupleres subject to reversion ([central anal sac] absent). The absence of the external anal pouch was infered to be the ancestral condition at the root of the feliformians, in agreement with Ewer (1973). The evolution of perineal glands also involved homoplastic events, but linear parsimony was unable to infer unambiguous patterns of character transformation. However, the picture obtained here is likely to be improved by a more accurate coding of these structures. For instance, due to the limited material available, Galidia was coded as "?" whereas perineal glands are known to be present (they are also present in Galidictis, another mongoose-like species of Madagascar; Pocock, 1915f; WCW, personal observation). Furthermore, the phenotypic expression of perineal glands is not as conserved as it may appear from our "practical" coding (e.g., in Arctogalidia, the glands are only present in males; glands do not show the same inner structure between the terrestrial civets and genets), and standardized, exhaustive anatomical investigations of these complex structures are greatly needed. Similar to the anal pouch, the absence of perineal glands was estimated as the ancestral condition in feliformians (proposed for the Order Carnivora by Flynn et al., 1988). From the mapping of these traits, two different developments of odoriferous system seem to have taken place during the evolution of Feliformia. The clade grouping Hyaenidae, Herpestidae and Malagasy taxa is characterized by two innovations concerning acquisition of external anal pouches (two types) but exhibit no perineal glands (Galidia excepted; see above). The clade Viverridae, its sister-group, is on the contrary characterized by the acquisition of three types of perineal glands (Poiana excepted), but does not show any structure similar to external anal pouches. Both odoriferous systems are involved in active biochemical communication (Ewer, 1973; Gorman et al., 1974; Kleiman, 1974; Wemmer, 1977; Roeder, 1980, 1984), and it is remarkable that sister lineages developed two completely different structures in order to fulfill a similar behavioral function.
The reconstructed evolutionary history of tympanic bulla conformation did not suggest homoplastic events. However, the reconstruction of ancestral traits proved ambiguous at the base of the ingroup. It thus remains unclear whether the condition met in Nandinia constitutes the most plesiomorphic state in feliformians (Hunt, 1987, 1998; Hunt and Tedford, 1993) or is an autapomorphy. On the other hand, it appeared clearly that the type of bulla characterizing the Felidae (type F) was derived from the symplesiomorphic conformation met in the Viverridae and Prionodontidae (type D). This is in conflict with the evolutionary hypothesis formulated by Hunt (1987), who considered that the bulla of Nandinia approximated an ancestral feliformian morphotype from which conformations met in fossil and extant lineages could be derived. Given the symplesiomorphic condition of type D, we recommend that bulla conformation should not be used as a diagnostic criterion for recognition of Viverridae in the fossil records (contrary to Hunt, 1974, 1987, 1991, 1998, 2001; also see Ivanoff, 2001). The assertion of Flynn et al. (1988), who suggested that the ear region of Cryptoprocta and Eupleres shared only plesiomorphies with the Viverridae (sensu stricto), cannot be evaluated because of ambiguous reconstruction.
Although this was not evidenced in all cases, we found that the evolutionary history of complex structures did involve homoplastic events (locomotion, anal and perineal glands), the causality of which remains obscure. Our evolutionary scenarios illustrate that complex structures are not forcedly submitted to irreversible evolutionary trends as has been also demonstrated for "specialized" morphologies in arthropods (Desutter-Grandcolas, 1997; D'Haese, 2000) and morphological evolution in a broader perspective (Seravin, 2001).
|
Conclusion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
Our study exemplified the need for considering independent (molecular) phylogenies in order to identify patterns of morphological homoplasy, to test the diagnostic value of complex structures, and to reconstruct evolutionary scenarios. It also demonstrated that node support needs to be estimated before pointing out the existence of conflicting topologies and drawing taxonomic conclusions. We illustrated the original, mostly nonadaptive patterns of convergence in viverrid-like taxa, and found noise conveyed by the morphological data set. Our investigations also demonstrated that noise, although difficult to characterize (notably via RI values), is troublesome in the context of phylogenetic reconstruction, at least when phylogenetic signal is heterogeneously distributed (but see Wenzel and Siddall, 1999). We thus urge caution in using only morphological characters for reconstructing phylogenetic relationships in feliformians. Instead of increasing the number of characters (Poe and Wiens, 2000) or invoking consensus methodologies based on topologies (i.e., supertrees; Bininda-Emonds et al., 1999), we recommend that a revision concerning the definition of and analytical procedures applied to morphological characters should be envisaged. A new analytical approach may prove benefitial to the study of morphological evolution among all mammalian lineages (e.g., Lovejoy et al., 1999).
|
Appendix 1. List and Description of the 349 Characters Constituting the Morphological Matrix |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
Skull
[Several features are described relative to each other. For these features, it is important to have the skull in an upright, horizontal position (the skull is considered horizontal, when the palate is held horizontally—NOT when the lower edge of the mandible is horizontal.]
Rostrum/Palatal Region
- Maxilla, infraorbital foramen (position of the anterior opening of the infraorbital canal): 0 = clearly anterior to the antorbital rim (i.e., long); 1 = slightly anterior to antorbital rim (i.e., short); 2 = ventral to the antorbital rim (i.e., very short). [The infraorbital canal serves for the passage of the infraorbital branch of the maxillary division of the trigeminal nerve (V) and the infraorbital branch of the maxillary artery.]
- Maxilla, infraorbital foramen, overall size: 0 = small; 1 = large.
- Maxilla, infraorbital foramen, shape: 0 = oval (dorsoventrally elongated); 1 = round.
- Maxilla, infraorbital foramen: 0 = dorsal and/or posterior to the P3-P4 area; 1 = dorsal to the P3-P2 area.
- Maxilla, infraorbital foramen, ventral border: 0 = posterior to dorsal border; 1 = ventral to dorsal border; 2 = anterior to dorsal border.
- Maxilla, infraorbital foramen, ventral border, relative position: 0 = dorsal to P3; 1 = dorsal to P4.
- Maxilla, infraorbital foramen, dorsal border: 0 = dorsal to P3; 1 = dorsal to P2; 2 = dorsal to P4.
- Maxilla, anterior alveolar foramen: 0 = within the infraorbital canal; 1 = within the orbit. [A small foramen that lies within the infraorbital canal, usually on the medial or ventral wall.]
- Maxilla, antorbital fossa: 0 = absent; 1 = present. [The antorbital fossa is on the rostrum and directly anteromedial to the lacrimal and the orbit. It serves for the insertion of the m. levator nasolabialis, which increases the diameter of the naris and lifts the apical portion of the upper lip.]
- Maxilla, nasolabialis fossa: 0 = absent; 1 = present. [This fossa lies on the ventral root of anterior surface of the zygoma and serves for the insertion of m. buccinator which returns food from the vestibule to the masticatory surface of the teeth.]
- Nasals, proportional dimensions: 0 = > 2x as long as wide; 1 = < 2x as long as wide. [Length measured at suture; width measured as widest point perpendicular to the anteriormost point of the nasal midline suture.]
- Nasals, posterior extension between frontals: 0 = extends posteriorly to the level of the posteriormost extension of the maxilla; 1 = does not extend posteriorly to the level of the posteriormost extension of the maxilla; 2 = extends posteriorly beyond the level of the posteriormost extension of the maxilla.
- Nasals, width: 0 = widest anteriorly and narrowing posteriorly; 1 = near parallel sides from maxilla/frontal suture; 2 = secondary widening at frontal/maxilla suture.
- Nasals and frontals, midline depression: 0 = present in both; 1 = present in nasals only; 2 = absent (in both).
- Maxilloturbinals: 0 = small; 1 = large and branching, excluding nasoturbinals from narial opening.
- Frontals, inflation anterior to the postorbital processes: 0 = absent; 1 = slight; 2 = large.
- Rostrum, dorsal profile: 0 = concave; 1 = straight (or nearly straight); 2 = convex.
- Rostrum, ‘trough’: 0 = rostrum extends anterior to maxilla/premaxilla palatal suture; 1 = rostrum is retracted slightly posterior to maxilla/premaxilla palatal suture. [This is caused by the retraction of the nasals, creating a shelf so that the dorsal (internal) portion of the premaxilla, which is normally not exposed, is exposed from a dorsal view of the skull.]
- Palate, anterior palatine foramina (= posterior palatine foramina of Crouch, 1969), relative position (de Beaumont, 1967; Qui, 1987; Wozencraft, 1989): 0 = adjacent to P4-M1; 1 = adjacent to P3; 2 = adjacent to P2; 3 = adjacent to P1. [This is the anterior opening of the palatine canal and serves for the passage of the palatine artery through the palatine canal (the posterior opening is in the ventral portion of the orbit).]
- Palatine/maxilla, anterior palatine foramina: 0 = located at palatine/maxilla suture; 1 = located well anterior in the maxilla.
- Palate, width, posterior to the canines: 0 = significantly wider than width measured at canines; 1 = nearly equal to width at canines. [Measured as the buccal width at the canines and the widest buccal width posterior to the canines.]
- Palate, tooth rows, lingual edge: 0 = nearly parallel; 1 = divergent.
- Palatal surface: 0 = flat or slightly concave; 1 = distinct longitudinal depression; 2 = extreme ‘vaulting.’
- Palatine/maxilla suture: 0 = medial or posterior to P4; 1 = medial to P3-P4; 2 = medial to P3; 3 = medial to P3-P2.
- Palatine, mesopterygoid palate: 0 = absent, or posterior border of the palate anterior to the posterior edge of the last teeth; 1 = present, short (width > length); 2 = present, long (width < length). [The mesopterygoid palatal region is here understood as the palatal region posterior to the last teeth, bridging the pterygoids. Length is measured as the posterior border of the palate to a line connecting the posterior edges of the last teeth. Width measured as the distance of the palate bridging the pterygoids.]
- Palatine, relative size (excluding mesopterygoid): 0 = midline length nearly equal in length to maxilla midline; 1 = midline length less than midline length of maxilla; 2 = midline length greater than midline length of maxilla.
- Palatine, relative size (including mesopterygoid): 0 = midline length nearly equal in length to maxilla midline; 1 = midline length less than midline length of maxilla; 2 = midline length greater than midline length of maxilla.
- Palatine, mesopterygoid region: 0 = flat/absent; 1 = slightly concave (depressed along midline).
- Palatine, mesopterygoid region: 0 = does not extend posteriorly to a point that would line up with the supraorbital processes; 1 = extends posteriorly to a point that would line up with the supraorbital processes; 2 = extends posteriorly to a point that would line up posterior with the supraorbital processes. [To understand this feature, the palate must be kept horizontal as a plane of reference.]
- Palatine, post palatal notch: 0 = postpalatal notch present; 1 = postpalatal notch absent. [This refers to the anterior emargination between the mesopterygoid region and the tooth row that in effect, causes the posterior border to be displaced anteriorly.]
- Palatine, extension of palate posterior to the tooth row: 0 = absent or vestigal; 1 = present, small (smaller than the length of the last tooth); 2 = present, large (longer than the length of the last tooth). [This refers to the portion of the palate, directly posterior (and in some cases posterolateral) to the tooth row and lateral to the mesopterygoid region.]
- Premaxilla, incisive foramina (= anterior palatine foramina of Crouch, 1969): 0 = oblong, long; 1 = oblong, short or rounded. [Serves for the passage of the nasopalatine branch of the trigeminal nerve (V), the nasal artery and the anterior septal branch of the major palatine.]
- Premaxilla/maxilla suture (midline) (Qiu, 1987): 0 = posterior to canines; 1 = medial to canines.
- Premaxilla, Steno's foramen (posterior edge): 0 = absent; 1 = medial to the posterior edge of the incisive foramina; 2 = posterior to the posterior edge of the incisive foramina. [Steno's foramen appears in some mammals as a foramen between the incisive foramina.]
- Vomer, attachment to palate: 0 = attached to anterior end of palatine; 1 = not attached to palatine (attached to maxilla).
- Orbit, anterior margin: 0 = dorsal to P4; 1 = dorsal to P3/P4; 2 = dorsal to P3; 3 = dorsal to P3/P2; 4 = posterior to P4. [Coding based on holding the palate horizontal/level.]
Orbital Region
- Alisphenoid, foramen rotundum: 0 = posteriorly separated from opening of sphenorbital fissure; 1 = closely adjacent to opening of sphenorbital fissure. [Serves for the passage of the maxillary division of the trigeminal nerve (V). The sphenorbital fissure serves for the passage of the oculomotor (III), trochlear (IV), abducens (VI), and trigeminal nerve (V).]
- Alisphenoid, foramen rotundum: 0 = lateral to sphenorbital fissure; 1 = ventral to sphenorbital fissure.
- Alisphenoid, (alar) canal (Flynn et al., 1988): 0 = present; 1 = absent. Serves for the passage of the maxillary artery. [Work by Novacek (1977), (1986) and Shoshani (1986) indicated that presence of the canal is primitive for this group. Flynn et al. (1988) coded absent as a synapomorphy for the Feliformia.]
- Alisphenoid/frontal muscle ridge: 0 = absent; 1 = present.
- Alisphenoid, optic foramen interorbital connection: 0 = partially blocked; 1 = clearly open; 2 = closed. [In some species, there is an open connection between the orbits through the optic foramen.]
- Frontal, supraorbital process: 0 = small process (maximum ‘tip to tip’ distance across supraorbitals less than 125% the width of the postorbital constriction); 1 = small ridge or protuberance; 2 = long process (greater than 125% the width of the postorbital constriction); 3 = long process, joins jugal; 4 = absent.
- Frontal-parietal suture (at midline): 0 = posterior to anterior edge of glenoid fossa; 1 = at level of anterior edge of glenoid fossa; 2 = anterior to anterior edge of glenoid fossa. [Baseline is a vertical line through the glenoid fossa with the palate held horizontal/level.]
- Frontal, temporal crest: 0 = present, single, well defined, ‘Y’-shaped; 1 = absent or small; 2 = present, double, nearly parallel; 3 = double, ‘lyre’-shaped. [Serves for the attachment of m. temporalis that raises the mandible.]
- Frontal, postorbital region (width): 0 = > interorbital region; 1 = approximately equal to interorbital region; 2 = < interorbital region.
- Frontal, interorbital inflation: 0 = absent; 1 = variable degrees of inflation.
- Frontal, postorbital inflation: 0 = absent; 1 = variable degrees of moderate inflation; 2 = greatly inflated.
- Postorbital constriction position: 0 = immediately posterior to supraorbital processes; 1 = considerably posterior to the supraorbital processes. [The postorbital constriction would be the point at which the interorbital region is the narrowest posterior to the supraorbital processes of the frontal.]
- Jugal: 0 = reaches lacrimal; 1 = does not reach lacrimal.
- Jugal/maxilla, posterioventral process: 0 = absent; 1 = present, weak; 2 = present, strong. [Serves for the attachment of the masseter.]
- Jugal, dorsal postorbital process: 0 = present, short process; 1 = small bump or ridge; 2 = absent; 3 = present, long; 4 = present, long, joining supraorbital process of frontal.
- Jugal, exposure on interorbital surface: 0 = large interior exposure; 1 = predominately confined to exterior surface.
- Lacrimal: 0 = large, extends ventral and dorsal to naso-lacrimal foramen; 1 = reduced, does not extend dorsal to naso-lacrimal foramen; 2 = reduced, does not extend ventral or dorsal to nasolacrimal foramen; 3 = large, extends dorsal, ventral and rostral to foramen.
- Lacrimal, inferior oblique muscle fossa: 0 = separate from nasolacrimal foramen, at lacrimal/maxilla suture; 1 = separate from nasolacrimal foramen, in maxilla. The fossa serves for the attachment of the m. obliquus ventralis which rotates the eye.
- Inferior oblique muscle fossa: 0 = single; 1 = double.
- Orbitosphenoid, anterior "process": 0 = long, directed anterodorsally; 1 = long, directed anteroventrally, nearly reaching sphenopalatine foramen; 2 = short, directed anteroventrally; 3 = absent.
- Orbitosphenoid, shape: 0 = reduced, oblong, restricted to area around optic foramen and the process anterior to it; 1 = irregular, with wing directly dorsal to optic foramen.
- Orbitosphenoid, sphenorbital fissure size: 0 = less than two times the size of the foramen rotundum; 1 = more than three times the size of the foramen rotundum; 2 = smaller than the size of the foramen rotundum.
- Maxilla, intraorbital wing: 0 = absent/vestigial; 1 = present (small).
- Maxilla, posterior border of zygomatic process: 0 = at plane posterior to P4-M1; 1 = at plane of P4-M1.
- Maxilla, anterior edge of orbit: 0 = posterior to P3-P4; 1 = at the plane of P3-P4; 2 = anterior to P3-P4.
- Maxilla, orbital process: 0 = expanded in orbital region and forming a horizontal shelf; 1 = not expanded in orbital region. [The maxilla expands outward in a horizontal plane in the rostral border of the orbit to form a triangular expansion ventral to the lacrimal.]
- Palatine, orbital process: 0 = not expanded in orbital region; 1 = greatly expanded in orbital region and forming a horizontal shelf.
- Palatine, sphenopalatine foramen: 0 = near equal in size to palatine canal; 1 = less than two times the size of the palatine canal; 2 = greater than twice as large as the palatine canal. [Serves for the passage of the sphenopalatine branch of the trigeminal nerve and the sphenopalatine artery.]
- Palatine, location of the sphenopalatine foramen (= orbitonasal foramen of Butler, 1948; sphenoid foramen of Novacek, 1986): 0 = dorsal to M2; 1 = dorsal to P4; 2 = dorsal to M1; 3 = anterior to P4; 4 = posterior to M2.
- Palatine, sphenopalatine foramen: 0 = confluent; 1 = closely adjacent to posterior opening of palatine canal; 2 = clearly separate. [Serves for the passage of the sphenopalatine branch of the trigeminal nerve (V) and the sphenopalatine artery.]
- Palatine, orbital wing: 0 = reaches lacrimal, broad contact; 1 = reaches lacrimal, narrow contact.
- Palatine, orbital exposure: 0 = straight or concave orbital wall; 1 = convex or inflated orbital wall.
- Pterygoid, epipterygoid processes: 0 = absent or weakly defined; 1 = present, forming narrow ridge; 2 = present, triangular process. [Serves for the attachment of m. pterygoid medialis which raises the mandible.]
- Pterygoid, hamular process: 0 = descends considerably ventral to the palate; 1 = approximately at the level of the palate.
Basicranial Region: Basioccipital-Basisphenoid-Alisphenoid
- Alisphenoid, foramen ovale (position): 0 = anterior to glenoid fossa; 1 = medial to glenoid fossa; 2 = posterior to glenoid fossa. [Comparisons made to the posteriormost edge of the glenoid fossa, not the process. Serves for the passage of the trigeminal nerve (V), internal maxillary artery and the medial meningeal artery.]
- Alisphenoid canal, anterior opening (foramen rotundum): 0 = posterior to pterygoids; 1 = dorsal to pterygoids; 2 = anterior to pterygoids. [Comparisons made while holding the palate horizontal/level.]
- Basioccipital, hypoglossal foramen (= condyloid foramen of Hunt, 1987; Flynn et al., 1988): 0 = distinctly separate from posterior lacerate foramen; 1 = closely adjacent; 2 = confluent with posterior lacerate foramen. [Located posteriomedial to the posterior lacerate foramen. State "1" was considered a synapomorphy for Feliformia by Flynn et al. (1988).]
- Basioccipital, posterior lacerate foramen (= jugular): 0 = large; 1 = small. [Serves for the passage of the internal jugular vein, and the glossopharyngeal (IX), vagus (X), and accessary (XI) nerves.]
- Basioccipital, muscle fossa: 0 = present; 1 = absent/weakly defined. [Serves for the insertion of m. rectus capitis ventralis, which draws the head ventral and depresses the snout.]
- Basioccipital, m. longus capitis rugosity: 0 = present; 1 = absent.
- Basioccipital, contact with entotympanic bullae: 0 = bassioccipital appressed and protrudes ventrally against bullae; 1 = bassioccipital does not protrude against bullae.
- Basioccipital width (between bullae at midpoint): 0 = narrower than pterygoid width; 1 = approximately equal to pterygoid width; 2 = wider than pterygoid width.
- Basisphenoid, medial lacerate foramen: 0 = covered by bullae; 1 = anteriomedial to bullae (e.g., Herpestes). [Serves for the passage of the ascending pharyngeal artery.]
- Basisphenoid, posterior opening of the vidian canal: 0 = present, posterior to pterygoids; 1 = present, medial to posterior edge of pterygoids; 2 = absent. [In the space between the posterior opening of the vidian canal and the bullae, the vidian nerve runs in a groove.]
Squamosal
- Squamosal, postglenoid foramen (Hough, 1948; Neff, 1983; Flynn et al., 1988): 0 = present, lateral to bullae on medial edge of the infrasquamosal fossa; 1 = vestigial/absent. [Flynn et al. (1988) listed ‘1’ as the synapomorphy for all Feliformia. Neff (1983) reported it as ‘reduced’ in viverrids and herpestids and absent in felids and hyaenids.]
- Squamosal, postglenoid process: 0 = closely appressed to auditory bulla; 1 = adjacent to auditory bulla; 2 = well separated from auditory bulla. [Reference point is posteriormost surface of glenoid fossa.]
- Squamosal, infrasquamosal fossa: 0 = present, narrow; 1 = present, wide; 2 = absent. [The infrasquamosal fossa is the portion of the ventral face of the squamosal that is anterolateral to the external auditory meatus.]
- Squamosal, lateral extension: 0 = both 1 and 2; 1 = expanded laterally, only dorsal to the external auditory meatus connecting the mastoid region with the zygomatic arch; 2 = expanded laterally and only dorsal to the mastoid; 3 = absent.
Ectotympanic
- Ectotympanic/entotympanic division: 0 = not externally visible; 1 = externally visible, without clear suture; 2 = externally visible, with clear suture. [In the viverrids, the point of contact between the two elements is marked by an inbending of the plates of bone. In the other feliforms, although one can detect the point of contact, it is because of structural differences.]
- Ectotympanic, ring, medial expansion (Hunt, 1987; Neff, 1983): 0 = moderate inflation and/or chambering; 1 = slight chambering or inflation; 2 = extremely large.
- Ectotympanic, ring, lateral expansion (external auditory meatus) (Hunt, 1987; Neff, 1983): 0 = a simple ‘C’ shaped ring with no significant lateral expansion; 1 = bowl; 2 = lateral outgrowths along rim nearly or completely meet along midline, forming flattened extension; 3 = meatal tube formed that scrolls anteriorly lateral to most lateral part of ectotympanic ring.
- Ectotympanic, ring, inclination: 0 = inclined; 1 = nearly vertical.
- Ectotympanic, inflation, form: 0 = medially expanded and inflated ‘C’ shape; 1 = slightly flattened, ‘fan’ shape; 2 = more extensive flattened ‘fan’ shape.
- Ectotympanic, direct ventral inflation to meatal tube: 0 = absent; 1 = present.
- Ectotympanic, septum (Hunt, 1987): 0 = complete (i.e., touches petrosal); 1 = partial; 2 = absent.
- Ectotympanic, styliform process: 0 = present; 1 = reduced; 2 = absent. [Anteromedial and ventral to the eustachian tube. Serves for the attachment of m. levator veli palatini.]
- Ectotympanic, eustachian tube: 0 = wide (nerve sulcus and tube widely separate); 1 = narrow (pterygoid nerve sulcus and eustachian tube closely adjacent).
- Ectotympanic, anteromedial process: 0 = present; 1 = absent. [The anteromedial process is on the anteromedial corner and dorsal to the eustachian tube.]
- Ectotympanic, external auditory meatus: 0 = level with nuchal crest; 1 = far forward of nuchal crest.
Entotympanic
- Entotympanic, relative inflation (Hunt, 1987): 0 = depth greater than ectotympanic ring; 1 = depth approximately equal to ectotympanic ring; 2 = entotympanic covered by ectotympanic, chambered; 3 = entotympanic not inflated.
- Entotympanic, ventral pseudoseptae: 0 = absent; 1 = present.
- Entotympanic, septum: 0 = anterior/posterior orientation; 1 = medial/lateral orientation; 2 = absent.
- Entotympanic, septum: 0 = absent; 1 = present. [Hunt (1987) contrasted Nandinia with no true internal septum to other feliforms. This refers to the entotympanic septum that developes from the ventral edge of the caudal entotympanic. Not homologous to the canid septum that develops from the dorsal edge (Hunt, 1987). Flynn, et al. (1988) listed the presence of a bilaminar septum as a synapomorphy for the feliforms (excluding Nandinia).]
- Entotympanic, carotid canal, posterior opening (Flynn et al., 1988): 0 = confluent with posterior lacerate foramen; 1 = clearly anterior to posterior lacerate foramen; 2 = absent (carotid canal absent).
- Entotympanic, course of the internal carotid: 0 = perbullar—between petrosal and basioccipital; 1 = perbullar—enclosed in entotympanic; 2 = transpromontorial groove in the entotympanic; 3 = transpromontorial. [Hunt (1987) coded herpestids and canids with a perbullar course of the internal carotid and felids, hyaenids, and viverrids with a transpromotorial course of the internal carotid.]
- Entotympanic ossification: 0 = ossified; 1 = unossified (i.e. Nandinia). [Flynn et al. (1988) listed ‘0’ as a synapomorphy for the Carnivora.]
- Entotympanic, rostral (Hunt, 1987): 0 = "athictic" entotympanic intervenes between ectotympanic and rostral entotympanic; 1 = "bradynothictic" entotympanic intervenes between ectotympanic and rostral entotympanic only in juveniles; 2 = "thictic" caudal entotympanic does not intervene between ectotympanic and rostral entotympanic. [Hunt (1987) established polarity on this feature on the ontogenetic growth sequence. He acknowledged that the athictic condition should be primitive based on outgroup comparison. Flynn et al. (1988) listed the presence of rostral entotympanic as a synapomorphy for the Carnivora.]
- Entotympanic, lateral expansion: 0 = no lateral expansion; 1 = lateral expansion predominately posterior to fenestra cochlea; 2 = lateral expansion extends directly lateral/anterolateral to fenestra cochlea.
- Entotympanic, anterior extension: 0 = large inflated anterior projection; 1 = small anterior extension; 2 = anterior extension restricted to tube.
- Entotympanic, antero-medial extension: 0 = entotympanic not expanded in antero-medial direction; 1 = extension to the level of the meatus; 2 = extension far forward the meatus.
- Entotympanic, carotid canal, anterior opening: 0 = horizontal; 1 = vertical.
Petrosal
- Petrosal, laterally directed mastoid pocket, middle ear cavity (Hunt, 1987): 0 = absent; 1 = present.
- Petrosal, ventral promontorial process (Hunt, 1987): 0 = vestigial/absent; 1 = present, small; 2 = present, large.
- Petrosal, posteromedial extension (Hunt, 1989): 0 = very large, with only a small space separating the petrosal from the occipital; 1 = large, with a moderately larger space separating the petrosal from the occipital; 2 = small, with a large space separating the petrosal from the occipital; 3 = very small, with an extremely large space separating the petrosal from the occipital.
- Petrosal, mastoid, petramastoid ridge: 0 = absent; 1 = present. [Origin of m. digastricus which opens the mouth. This ridge of mastoid connects the paroccipital process with the mastoid process.]
- Petrosal, inferior petrosal sinus (petrobasilar canal): 0 = large, open; 1 = small. [Closely related to the articulation of the petrosal to the basioccipital.]
- Petrosal, subarcuate (floccular) fossa: 0 = deep; 1 = shallow. [The subarcuate (floccular) fossa is a large fossa formed within the dorsal loop of the superior semicircular canal and houses the paraflocculus of the cerebellum.]
- Petrosal, fenestra cochleae (round window): 0 = approximately equal in size to oval window; 1 = considerably greater in size than oval window.
- Petrosal, basal whorl of cochlea: 0 = not distinctly swollen; 1 = distinctly swollen.
- Petrosal, mastoid, exposed ventrolateral process: 0 = large ventrally directed mastoid process, pendulant; 1 = pendulant, process absent, large deltoid exposure of mastoid, 2 = process absent, small deltoid exposure of mastoid; 3 = process nearly excluded from ventrolateral exterior surface.
- Petrosal, mastoid, occipital exposure: 0 = present; 1 = absent.
- Petrosal, mastoid, mastoid foramen: 0 = present; 1 = absent.
- Dorsal pseudoseptae: 0 = absent; 1 = present.
- Petrosal, fossa muscularis major (tensor tympani fossa): 0 = present, deep; 1 = present, shallow. [Origin of m. tensor tympani that tenses the tympanic membrane.]
- Petrosal, promontorium (Hunt, 1989): 0 = smooth; 1 = faceted.
Paraoccipital
- Paroccipital process (Neff, 1983; Hunt, 1987; Flynn et al., 1988): 0 = long, cupped around posterior edge of entotympanic and extends ventrally; 1 = long, separate, protrudes posteriorly; 2 = short, cupped around posterior edge of entotympanic and does not protrude ventrally; 3 = very short, does not cover posterior surface; 4 = very short, not cupped around posterior edge of entotympanic, and protrudes posteriorly but does extend ventrally. [Origin of m. jugulohyoideus that draws the hyoid caudad.]
- Paroccipital process, width (Petter, 1974; Veron, 1995): 0 = paroccipital process narrow and not covering all the posterior surface of the mastoid; 1 = paraoccipital process very wide and covering all the posterior surface of the mastoid, so that the mastoid can not be seen in a caudal view of the skull.
Ossicles
- Ossicles, malleus, anterior (long) process: 0 = present, long; 1 = present, short.
- Ossicles, malleus, processus gracilis and anterior lamina: 0 = well developed; 1 = reduced.
- Ossicles, malleus, head: 0 = 2 facets; 1 = 3 facets.
- Ossicles, malleus, head articulation surface, upper facet: 0 = convex; 1 = flat.
Basicranium (Other Parts)
- Presphenoid, shape: 0 = long narrow ridge; 1 = wide, flat.
- Vagina processus hyoideus: 0 = confluent with stylomastoid foramen (giving the appearance of one foramen); 1 = slightly separate from stylomastoid foramen.
Auditory Bullae—General
- External auditory meatus, dorsal wall: 0 = ectotympanic; 1 = squamosal.
- Internal carotid artery (Neff, 1983): 0 = large; 1 = small; 2 = vestigial/absent.
- Arterial circulation, intracranial rete (Bugge, 1978): 0 = absent; 1 = present.
- Arterial circulation, major a1 arterial shunt (Bugge, 1978): 0 = present; 1 = absent.
- Arterial circulation, major a4 arterial shunt (Bugge, 1978; Neff, 1983; Wozencraft, 1989): 0 = absent; 1 = present, intracranial rete.
- Arterial circulation, major anastomosis ‘X’ (Bugge, 1978; Wozencraft, 1989): 0 = present; 1 = vestigial/absent.
- Arterial circulation, major anastomosis ‘Y’ (Bugge, 1978; Wozencraft, 1989): 0 = absent; 1 = present.
- Arterial circulation, major a2 arterial shunt (Bugge, 1978; Wozencraft, 1989): 0 = absent; 1 = present, intracranial rete.
- M. sternomastoideus and m. cleidomastoideus fossa (location): 0 = principally on mastoid; 1 = on squamosal and mastoid; 2 = principally on squamosal; 3 = principally on entotympanic bullae.
- Sternomastoideus and cleidomastoideus fossa: 0 = vestigial/reduced; 1 = pronounced.
- Bullar configuration (Hunt, 1987): 0 = anterior migration of entotympanic beneath ectotympanic; 1 = no change during ontogeny; 2 = restriction of entotympanic to posterior auditory region, septum bullae transverse or inclined; 3 = greater restriction of entotympanic to posterior auditory region.
Cranial
- Exoccipital, condyloid canal (occipital venous sinus): 0 = large; 1 = small/vestigial. [Serves for the passage of the occipital venous sinus.]
- Supraoccipital: 0 = not inflated; 1 = inflated.
- Condyles, emarginated causing the condyle to appear slightly lobed: 0 = absent; 1 = present.
- Parietal, sagittal crest (adult males): 0 = large, single; 1 = small, single; 2 = small double; 3 = absent; 4 = large, double.
- Parietal/supraoccipital, lambdoidal crest: 0 = not well produced dorsally or laterally beyond braincase; 1 = well produced dorsally beyond braincase; 2 = well produced laterally and dorsally beyond braincase.
- Parietal, sagittal crest: 0 = present, does not extend beyond lambdoidal crest; 1 = present, extends beyond lambdoial crest.
- Parietal/supraoccipital, lambdoidal crest, caudal view: 0 = narrow ‘V’; 1 = narrow ‘U’; 2 = broad ‘U.’
Dentition
- I2 position (posterior edge): 0 = in line with I1 and I3; 1 = set back.
- I3, form (cross section): 0 = posterior broadening of crown, oblong; 1 = incisiform; 2 = caniniform.
- I3, size: 0 = considerably larger than I1 or I2; 1 = slightly larger than I1 or I2; 2 = equal in size to I1 or I2.
- I1-I3, projection: 0 = extends upward; 1 = procumbent (less than 45 degrees).
- I1-I3 formation: 0 = in a gently curved line; 1 = in a straight line; 2 = sharply curved line (I3 laterally placed).
- I1-I3 formation: 0 = incisors close together; 1 = incisors separated from each other.
- I3/C1 occlusion: 0 = present, small wear facet; 1 = present, large wear facet; 2 = absent.
- C1, size: 0 = large (1.5 to 2.5 times as high as I3); 1 = small (equal in size, or less than 1.5 times as high as I3).
- C1, flanges: 0 = posterior flange present; 1 = absent; 2 = anterior and posterior flanges present.
- C1, shape: 0 = conical, recurved; 1 = dagger like, straight.
- P1: 0 = present; 1 = present in young adults, but lost in many old individuals; 2 = absent.
- P1/1, relative size: 0 = small, less than half the size of P2/2 or absent; 1 = small, < P2/2, but more than state "0"; 2 = approximately equal to P2/2.
- P1, number of roots: 0 = 1 root or tooth absent; 1 = 2 roots, completely or partially fused.
- P1: 0 = present; 1 = present in young adults, but lost in many old individuals; 2 = absent.
- P2, number of roots: 0 = 2; 1 = 1.
- P2, roots: 0 = 2; 1 = 1; 2 = absent.
- P3, number or roots: 0 = 2; 1 = 1; 2 = 3.
- P3, posterolingual cingulum: 0 = present, reduced; 1 = present, broad; 2 = vestigial/absent.
- P3, lingual cusp: 0 = absent; 1 = present.
- P3/3, size (when viewed from the buccal aspect): 0 = nearly equal to P2/2; 1 = considerably larger than P2/2.
- P3, roots: 0 = 2; 1 = 1.
- P3, talonid: 0 = absent; 1 = present.
- P4, number of cusps: 0 = > 1; 1 = 1.
- P4, roots: 0 = 3 separate roots; 1 = 2 roots (one anterior, one posterior).
- P4, parastyle: 0 = absent/vestigial; 1 = present, small (less than size of protocone); 2 = present, large (near equal to protocone); 3 = present, larger than protocone.
- P4, metacone or metastylar blade: 0 = present; 1 = vestigial; 2 = absent.
- P4, lingual lobe: 0 = occupied by protocone which is less than paracone; 1 = occupied by extremely small protocone which is much less than paracone; 2 = occupied by large protocone which is near equal to paracone. 3 = vestigial/absent.
- P4, talon: 0 = vestigial/absent; 1 = present.
- P4, metastyle blade: 0 = long, with notch reduced to slit; 1 = short, with V shaped notch; 2 = absent/vestigial.
- P4, posterolingual cingulum: 0 = present; 1 = vestigial; 2 = absent.
- P4, talonid: 0 = absent/vestigial; 1 = present.
- P4, posterior accessory cusps: 0 = 2 cusps, linear; 1 = 1 cusp; 2 = no cusps; 3 = 2 or more cusps, nonlinear. [Flynn and Galiano (1982) considered P4 well developed with two accessory cusps plesiomorphic for the Carnivora and the lost of accessory cusps derived.]
- Carnassial shear, P4/M1 (well defined vertical facet) (Flynn and Galiano, 1982): 0 = present; 1 = absent.
- P4-M1 carnassial embrasure pit: 0 = present; 1 = absent.
- M1, shape: 0 = rhomboidal; 1 = triangular; 2 = round; 3 = rectangular (length larger than width).
- M1, roots: 0 = 3; 1 = 2; 2 = 1.
- M1, hypocone: 0 = present; 1 = absent.
- M1, buccal lobes: 0 = asymmetrical or absent; 1 = symmetrical.
- M1, trigonid (or equivalent): 0 = width < length; 1 = width > length; 2 = width = length; 3 = vestigial/absent.
- M1, trigonid: 0 = present, considerably higher than talonid; 1 = present, near equal in height to talonid; 2 = vestigial; 3 = absent.
- M1, talonid: 0 = present, length < length of trigonid; 1 = vestigial or absent; 2 = present, length > length of trigonid; 3 = length near equal length of trigonid.
- M1, metaconid: 0 = large; 1 = considerably smaller than paraconid; 2 = absent.
- M1, protoconid: 0 = present; 1 = absent.
- M1, entoconid: 0 = present; 1 = absent.
- M1, hypoconid: 0 = present; 1 = absent.
- M1, hypoconulid: 0 = absent; 1 = present.
- M1, small cusp between hypoconid and protoconid: 0 = absent; 1 = present.
- M2, paracone and metacone: 0 = metacone and paracone present, paracone larger; 1 = metacone absent; 2 = tooth without distinguishable cusps; ? = tooth absent.
- M2: 0 = present; 1 = absent.
- M2, hypocone: 0 = absent; 1 = present; ? = tooth absent.
- M2, roots: 0 = 3; 1 = 2; 2 = 1; ? = tooth absent.
- M2, shape: 0 = rhomboidal; 1 = triangular; 2 = round; ? = tooth absent.
- M2: 0 = present; 1 = absent.
- M2, roots: 0 = > 1; 1 = 1; ? = tooth absent.
- M2, cusps: 0 = tooth with cusps but no clear trigonid or talonid; 1 = trigonid present; 2 = tooth without clearly defined cusps; ? = tooth absent.
- M2/ M2 relative size: 0 = considerably smaller than M1/m1; 1 = slightly smaller than M1/M1; ? = tooth absent.
- M3: 0 = present; 1 = absent.
- Upper molars, cingulum: 0 = present, extremely large posteriolingual extension; 1 = absent, if present, not complete; 2 = present, complete around the base of the protocone.
- Buccal cingulum, upper molars: 0 = small/not developed; 1 = enlarged.
- Diastema, I2-I3: 0 = absent; 1 = present.
- Diastema, C1-P1 (or to first postcanine tooth): 0 = absent; 1 = present.
- Diastema, P2-P3 (or second and third postcanine teeth): 0 = small to absent; 1 = large.
- Diastema, P1-P2 (or first two postcanine teeth): 0 = small to absent; 1 = large.
- Mandible, angular process: 0 = present, strong; 1 = present, weak.
- Mandible, mandibular symphysis: 0 = strong (well fused); 1 = weak (not well fused).
- Mandible, mandibular condyle: 0 = considerably elevated above alveolar line; 1 = slightly above alveolar line; 2 = at alveolar line; 3 = below alveolar line. [Alveolar line defined as a line connecting the anterior edge of the canine alveolus with the posterior edge of the last tooth.]
- Mandible, medial shelf of angular process: 0 = present, small ridge; 1 = present, wide shelf; 2 = absent.
- dP3, number of accessory posterior cusps: 0 = 1; 1 = 2; 2 = 3; 3 = 0.
- dP3 talonid: 0 = present; 1 = absent.
- dP3, lingual cusp: 0 = present, large; 1 = absent; 2 = present, small.
- dP4, buccal lobes: 0 = slightly asymmetrical; 1 = strongly asymmetrical; 2 = symmetrical.
- dP3, accessory lingual cusp: 0 = absent; 1 = present.
- Mental foramina: 0 = more than one; 1 = one only.
- P4, anterior accessory cusp: 0 = absent; 1 = smaller than posterior; 2 = about equal to posterior; 3 = larger than posterior.
Postcranium
Vertebral Column
- Atlas, transverse process, posterior extension: 0 = to the level of the caudal articular surface; 1 = slightly posterior to the caudal articular surface; 2 = considerably posterior to the caudal articular surface. [M. levator scapulae attaches to the transverse process and serves to pull the scapula forward.]
- Atlas, alar and atlantal (intervertebral) foramen: 0 = alar foramen not complete, notch present; 1 = alar foramen complete, widely separate from atlantal foramen; 2 = alar foramen complete, closely adjacent to atlantal foramen in common groove.
- Atlas, ventral arch, length: 0 = equal to or shorter than dens of axis; 1 = longer than dens of axis.
- Atlas, transverse foramen (vertebral artery), posterior opening: 0 = on dorsal surface; 1 = caudal.
- Axis, cranial articular processes and ventral surface of dens (odontoid) process: 0 = discrete surfaces; 1 = one continuous surface.
- Axis, spinous process: 0 = with a relatively flat dorsal profile; 1 = with a convex profile.
- Axis, caudal spine of spinous process: 0 = absent; 1 = present, short; 2 = present, long.
- Axis, ventral ‘keel’: 0 = present, small; 1 = absent/vestigial; 2 = present, large.
- Cervical vertebrae, ventral keel on vertebrae 6 and 7: 0 = vestigial/absent; 1 = present.
- Sacral vertebrae, spinous processes: 0 = separate processes; 1 = fused together into one straight ‘keel.’
- Tail: 0 = much shorter than head-body length (15–40%); 1 = shorter than head-body length (40–85%); 2 = approximately equal to head-body length (85–105%); 3 = much longer than head-body length (> 105%); 4 = vestigial (0–15%).
Thoracic Skeleton
- Scapula, teres major process (axillary border) (Taylor, 1974): 0 = present, small perpendicular plane to scapular plane; 1 = present, small, on plane of scapula; 2 = present, large perpendicular plane to scapular plane. [A large teres major process is a synapomorphy for Caniformia (Flynn et al., 1988).]
- Scapula, metacromion/acromion processes: 0 = acromion process larger; 1 = processes approximately equal in size; 2 = metacromion process larger. [The m. acromiodeltoideus attaches to the acromion flexor and outer rotator of the humerus.]
- Scapula, metacromion process (shape): 0 = deltoid; 1 = square.
- Scapula, coracoid process: 0 = present, well developed; 1 = present, small; 2 = vestigial/absent.
- Scapula, supraspinus fossa (cranial border): 0 = larger than infraspinous fossa; 1 = near equal in size.
- Scapula, overall shape: 0 = rectangular; 1 = rounded; 2 = triangular.
- Thoracic vertebrae, number: 0 = 13; 1 = 14; 2 = 15; 3 = 12.
- Humerus, supracondylar (= entepicondyloid) foramen: 0 = absent; 1 = present.
- Humerus, pectoral crest: 0 = slender with little development of crest; 1 = absent; 2 = robust, with crest developed toward midpoint of distal end. [Pectoral ridge starts proximally at the anterior edge of the greater tuberosity of the humerus; serves for the attachment of the m. pectoralis major.]
- Humerus, deltoid crest: 0 = robust, with crest developed towards midpoint of distal end; 1 = slender with little development of crest; 2 = absent. [Deltoid crest starts proximally at the infraspinous facet of the greater tuberosity of the humerus (lateral side); serves for the attachment of the acromiodeltoid.]
- Humerus, olecranon fossa: 0 = covered by membrane or open; 1 = ossified. [Allows for greater arc movement of the antebrachium.]
- Humerus, distal condyle: 0 = trochlea distal extension greater than capitulum; 1 = trochlea distal extension approximately equal to capitulum.
- Humerus, radial and coronoid fossas: 0 = confluent; 1 = distinct.
- Humerus, greater tuberosity, proximal end: 0 = supraspinatus fossa large with robust deltoid anterior process (near equal in size to articular surface); 1 = supraspinatus fossa small, restricted to narrow ridge, anterior process small (considerably smaller than articular surface).
- Humerus, greater tuberosity, proximal end: 0 = tall, extends beyond articular surface; 1 = very tall, extends considerably beyond articular surface; 2 = short, extends only slightly beyond articular surface; 3 = equal, equal in height to articular surface.
- Humerus, distal supracondyloid crest: 0 = absent/vestigial; 1 = small; 2 = well developed, flares out. [Serves for the insertion of the extensor carpi radialis longus and brachioradialis.]
- Humerus, cross section: 0 = oval; 1 = round; 2 = flattened.
- Humerus, deltoid and pectoral crest confluence: 0 = proximal; 1 = midpoint.
- Ulna, olecranon process, medial and lateral ridges: 0 = lateral ridge larger; 1 = approximately equal; 2 = vestigial/absent.
- Ulna, olecranon process, medial fossa with ventral/medial flange: 0 = vestigial/absent; 1 = present, small; 2 = present, large.
- Ulna, distal lateral shelf: 0 = absent; 1 = present. [Serves for the attachment of the anconeus.]
- Ulna, olecranon process: 0 = large ventral process—squared off outline; 1 = ventral process, vestigial/absent.
- Ulna, olecranon, inclination (Taylor, 1974): 0 = posterior; 1 = perpendicular; 2 = anterior.
- Radius, cross section: 0 = flattened; 1 = oval; 2 = round.
- Radius, tubercle: 0 = present, small; 1 = vestigial/absent; 2 = present, large.
- Radius, distal expansion: 0 = vestigial/absent; 1 = present.
- Radius, head—articular surface: 0 = distinct ridge; 1 = without distinct ridge.
- Manus, digit 1 (hallux): 0 = present, small (< 0.5 digit 2); 1 = vestigial/absent; 2 = present, normal (approximately = 0.5 digit 2); 3 = present, large (> 0.5 digit 2).
Pelvic Skeleton
- Ilium, ventral iliac spine (King, 1966): 0 = not everted or extended; 1 = everted, extended anterolaterally.
- Pelvis, overall shape: 0 = triangular (thick fusion); 1 = rectangular shape (thin fusion); 2 = rhomboidal.
- Femur, trochanteric fossa: 0 = restricted to ‘V’ shape above lesser trochanter; 1 = expanded medially from lesser trochanter; 2 = expanded distally from lesser trochanter.
- Femur, lesser trochanter: 0 = present, small; 1 = present, large.
- Femur, lesser trochanter, position: 0 = intermediate position aligned with lateral edge of head; 1 = medial position, directly below head; 2 = lateral position, lateral to lateral edge of head. [Lesser trochanter aligns with lateral edge of head articular surface.]
- Femur, cross-section: 0 = oval; 1 = round; 2 = flattened.
- Tibia, cross-section: 0 = round; 1 = oval; 2 = triangular.
- Tibia-fibula, proximal fusion: 0 = absent; 1 = present.
Pes
- Calcaneum, sustentaculum tali (Robinette and Stains, 1970): 0 = without secondary shelf; 1 = small secondary shelf.
- Calcaneum, calcaneal tubercle or process: 0 = present, symmetrical; 1 = present, asymmetrical.
- Calcaneum, medial tuberosity on calcaneal tubercle: 0 = present, larger than lateral tuberosity; 1 = vestigial/absent; 2 = approximately equal to lateral tuberosity.
- Calcaneum, peroneal tubercle: 0 = distal to astragular tubercle, small; 1 = medial to astragular tubercle.
- Calcaneum, medial articular surface: 0 = small; 1 = large, extending to cuboid surface.
- Calcaneum, cuboid surface: 0 = perpendicular to axis; 1 = slight medio-lateral slope; 2 = sharp medio-lateral slope; 3 = dorsal slope.
- Calcaneum, trochlear process: 0 = vestigial/absent; 1 = small; 2 = large.
- Calcaneum, posterior projection off of cuboid surface: 0 = vestigial/absent; 1 = present.
- Pes, digit 1 (pollex): 0 = vestigial/absent; 1 = present, small (< 0.5 digit 2); 2 = present, normal (approximately = 0.5 digit 2); 3 = present, large (> 0.5 digit 2).
- Baculum: 0 = long, stylized; 1 = small, simple; 2 = vestigial, absent.
External/ Soft Anatomy
- Rhinarium, philtrum: 0 = present, narrow; 1 = present, wide; 2 = absent. [Externally visible rhinarial tract that connects the rhinarium to the lower lip.]
- Rhinarium, groove: 0 = present, extends dorsal to a point medial to nares; 1 = present, only ventral to nares; 2 = present, extends dorsal and onto dorsal surface; 3 = absent/vestigial.
- Rhinarium, external nares: 0 = extends onto lateral surface (not visible on dorsal surface); 1 = extends onto dorsal surface; 2 = restricted to dorsal surface.
- Rhinarium, dorsal exposure: 0 = approximately equal to anterior exposure; 1 = much less than anterior exposure; 2 = longer than anterior exposure.
- Rhinarium, profile: 0 = perpendicular to axis; 1 = convex; 2 = concave.
- External pinnae, aural posteroexternal bursa: 0 = present; 1 = absent.
- External pinnae, posterior, subbursa lobe: 0 = absent; 1 = present (a posterior external lobe that projects ventral to the ear bursa).
- External pinnae: 0 = present, angular or pointed; 1 = present, rounded.
- Ear bursa: 0 = posterior flap of bursa behind the rim of the pinna; 1 = ear bursa absent; 2 = posterior flap of the bursa continuous with the rim of the pinnae.
- Perineal scent glands, males: 0 = absent; 1 = present, restricted to area between penis and anal region; 2 = incircles base of penis like a collar; 3 = present, extends anterior to penis.
- Perineal glands: 0 = absent; 1 = present in both sexes or female only.
- Perineal glands, male: 0 = gland absent; 1 = hair between gland and anus; 2 = hairless between gland and anus.
- Scrotum, position: 0 = closely associated with penis, common scrotal sac; 1 = widely separated from penis, common scrotal sac; 2 = widely separated from penis, separated scrotal sacs.
- Mammae: 0 = 8; 1 = 6; 2 = 4; 3 = 2.
- Perineal scent glands, females: 0 = absent; 1 = centered around the vulva; 2 = present, restricted to the area between the vulva and anus; 3 = extends anterior to the vulva.
- Perineal glands, female: 0 = gland absent; 1 = hairless between gland and anus; 2 = hair between gland and anus.
- Anal glands: 0 = simple, open into small anal sac; 1 = open into anal pouch.
- Feet, interdigital pads (manus and pes): 0 = 3; 1 = 4.
- Feet, interdigital webbing of external foot pads: 0 = present, sparsely furred; 1 = present, well furred; 2 = present, without hair; 3 = absent.
- Feet, interdigital pads: 0 = completely fused together; 1 = partially fused together, shallow grooves; 2 = closely adjacent to each other, deep grooves; 3 = well separated from each other.
- Feet (manus), carpal (palmar) surface, central depression: 0 = absent; 1 = present, with hair; 2 = present, without hair.
- Feet (manus), metacarpal pad(s): 0 = absent; 1 = hypothenar pad vestigial; 2 = hypothenar pad large, thenar pad vestigial; 3 = hypothenar and thenar pads large and well developed; 4 = metacarpal pads very large and completely fused.
- Feet (manus), carpal pollical pad: 0 = absent; 1 = separated from other carpal pads by bare skin; 2 = adjacent to other carpal pads.
- Feet (manus), carpal pollical pad, size: 0 = larger than digital pads; 1 = about equal to digital pads; ? = absent.
- Feet (manus), metacarpal pad(s), size: 0 = vestigial; 1 = smaller than carpal pads; 2 = large, about equal to carpal pads; 3 = larger than carpal pads; ? = absent.
- External pinnae, anterior surface: 0 = lighter than ground color; 1 = same as ground color; 2 = darker than ground color; 3 = white rim.
- Feet (pes), plantar surface, central depression: 0 = absent; 1 = present, without hair; 2 = elongate; 3 = elongate and with papillae confined to heel.
- Feet (pes), metatarsal pad(s): 0 = absent/vestigial; 1 = connected to tarsal plantar pads (no clear separation); 2 = partially connect to plantar pads; 3 = not connected.
- Feet (pes), metatarsal surface (posterior to metatarsal pads): 0 = well furred; 1 = without hair.
- Feet (pes), metatarsal pad(s): 0 = separated from tarsal pads by hair; 1 = adjacent to lateral tarsal pad and hallical pad; 2 = separated from tarsal pads by bare skin; ? = absent.
- Feet (pes), tarsal hallical pad: 0 = absent; 1 = separated from other tarsal pads by hair; 2 = separated from other tarsal pads by bare skin; 3 = adjacent to other tarsal pads.
- Feet (pes), tarsal hallical pad, size: 0 = small than digital pads; 1 = about equal to digital pads; 2 = larger than digital pads; ? = absent.
- Feet (pes), metatarsal pad(s), size: 0 = vestigial; 1 = small; 2 = large; 3 = larger than tarsal pads; ? = absent.
- Head, dark face mask: 0 = absent; 1 = present.
- Head, interoccular (between) eye patches: 0 = absent; 1 = present.
- Head, rhinarial patch: 0 = present, lighter than ground color; 1 = absent.
- Head, suboccular eye patches: 0 = present, indistinct; 1 = absent; 2 = present, distinct.
- Head, ‘rostral line’: 0 = absent; 1 = present, dark; 2 = present, light.
- Head, supraoccular eye patches: 0 = present, indistinct; 1 = absent; 2 = present, distinct.
- Neck, dorsal surface: 0 = same as ground color; 1 = darker than ground color; 2 = clearer than ground color.
- Nuchal stripes: 0 = absent; 1 = thin; 2 = large.
- Neck, dorsal hair direction: 0 = posteriorly directed; 1 = anteriorly directed.
- Neck, sides: 0 = uniform color; 1 = spotted; 2 = laterally directed stripes.
- Body, spots: 0 = no pattern, monocolored; 1 = present, no pattern; 2 = arranged into longitudinal lines; 3 = merged into broad longitudinal lines; 4 = merged into transverse lines.
- Body, pale shoulder spots: 0 = absent; 1 = present.
- Body, dorsal crest (long dorsal hairs on mid-dorsal line): 0 = absent; 1 = present.
- Body, mid-dorsal stripe: 0 = absent; 1 = present, extends from head to base of tail; 2 = extends from shoulders to base of tail.
- Body, mid-dorsal stripe, color: 0 = present, same color as body spots; 1 = present, darker than body spots; 2 = present, lighter than body spots; ? = absent.
- Body, ventral surface: 0 = lighter than ground color; 1 = same as ground color; 2 = lighter and spotted.
- Feet: 0 = same as ground color; 1 = spotted; 2 = dark.
- Legs: 0 = same as ground color; 1 = spotted; 2 = dark.
- Tail, tip: 0 = darker than ground color; 1 = same as ground color; 2 = lighter than ground color.
- Tail, hair length: 1 = short guard hair; 0 = long guard hair.
- Tail, dorsal line: 0 = absent; 1 = present.
- External pinnae, posterior surface: 0 = same as ground color; 1 = lighter than ground color; 2 = darker than ground color; 3 = darker than ground color with white rim; 4 = darker than ground color with white spot.
- External pinnae, tuft: 0 = absent; 1 = present.
- Tail, fat storage: 0 = absent; 1 = present.
- Tail, mobility: 0 = not prehensile; 1 = prehensile.
- Cowper's (bulbourethral) gland: 0 = absent/vestigial; 1 = present.
- M. accessorius pedis: 0 = present; 1 = absent.
- M. rhomboideus major: 0 = attaches to vertebrae; 1 = attaches to occiput.
- M. supinator longus (Windle and Parsons, 1897): 0 = single; 1 = double.
- M. semitendinosus, caudal head (Windle and Parsons, 1897): 0 = absent; 1 = present.
- M. soleus (Windle and Parsons, 1897): 0 = absent; 1 = present.
- Cruciate sulcus (Radinsky, 1971, 1975; Carlsson, 1900): 0 = present, large; 1 = present, small; 2 = absent.
- Postlateral sulcus does not overlap lateral sulcus (Radinsky, 1975): 0 = yes; 1 = no (overlap).
- Presylvian sulcus (Radinsky, 1975): 0 = present; 1 = absent.
- Suprasylvian sulcus (Radinsky, 1975): 0 = complete arch; 1 = incomplete arch.
- Vibrissae, inter-ramal tuft: 0 = present; 1 = absent.
- Pupil, horizontally elongated: 0 = no; 1 = yes.
- Marker chromosome (Wurster, 1969): 0 = present; 1 = absent.
|
Acknowledgments |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
We wish to thank the following people and institutions for providing access to collection specimens: Michel Tranier and Jacques Cuisin, Muséum National d'Histoire Naturelle, Paris, France (MNHN); Marc Colyn, Station Biologique de Paimpont, Université de Rennes 1, France (SBPUR1); Rainer Hutterer and Gustav Peters, Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn, Germany (ZFMK); Rennata Angermann, Manfred Ade and Irene Thomas, Museum für Naturkunde (Humboldt University), Berlin, Germany (MFNB); Dieter Kock, Forschungsinstitut Senckenberg, Frankfort, Germany; Paula Jenkins and Daphne Hills, Natural History Museum, London, U.K. (BMNH); Wim Van Neer and Wim Wendelin, Musée Royal d'Afrique Centrale, Tervuren, Belgium (MRAC); Georges Langlet, Institut royal des Sciences naturelles de Belgique, Bruxelles, Belgium (IRSNB); Chris Smeenk, National Museum of Natural History, Leiden, Holland (RMNH); Peter John Taylor, Durban Natural Science Museum, S.A. (DNSM); Judith Masters, the Natal Museum, Pietermaritzburg, S.A. (NM); Denise Drinkrow, the South African Museum, Cape Town, S.A. (SAM); Bruce Patterson and Larry Heaney, Field Museum of Natural History, Chicago, U.S.A. (FMNH); Frank Iwen, University of Wisconsin Zoological Museum, Madison, U.S.A. (UWZM); Robert Hoffmann and Linda Gordon, National Museum of Natural History, Smithsonian Institution, Washington, U.S.A. (NMNH); Thor Holmes and Larry Martin, Museum of Natural History, University of Kansas, Lawrence, U.S.A. (KMNH); Maria Rutzmoser, Museum of Comparative Zoology, Harvard, U.S.A. (MCZ); Guy Musser, Sidney Anderson and André Wyss, American Museum of Natural History, New York, U.S.A. (AMNH); Mark Taylor and Corey Goldmann, Royal Ontario Museum, Toronto, Canada (ROM). WCW would like to thank the following people for their support and assistance: Norman Bridges, Mary Boise, Nelson Worden, Bethel College; Lothar Schlawe, Zoologisches Institut der Freien Universitat Berlin; Annalisa Berta, San Diego State University. PG and GV received financial support from the Bioresource Program (European Union's Training and Mobility of Researchers Program) for visiting the BMNH collections in 2001. WCW's financial support was provided by the University of Kansas, Smithsonian Institution, and Bethel College. We are grateful to Olaf Bininda-Emonds, John Finarelli, Katie Weakland, Catherine Pecoraro, and one anonymous reviewer for greatly improving the quality of the early draft of this manuscript.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionConclusionAppendix 1. List and...AcknowledgmentsReferences |
|---|
-
Albignac R. Faune de Madagascar. In: Mammifères Carnivores. (1973) Paris: ORSTOM—CNRS.
Alvarez Y., Javier Juste B., Tabares E., Garrido-Pertierra A., Ibanez C., Bautista J. M. Molecular phylogeny and morphological homoplasy in Fruitbats. Mol. Biol. Evol. (1999) 16:1061–1067.[Abstract]
Baker R. H., DeSalle R. Multiple sources of character information and the phylogeny of Hawaiian drosophilids. Syst. Biol. (1997) 46:654–673.
Baker R. H., Yu X., DeSalle R. Assessing the relative contribution of molecular and morphological characters in simultaneous analysis trees. Mol. Phylogen. Evol. (1998) 9:427–436.[CrossRef][Web of Science][Medline]
Bauer D. F. Constructing confidence sets using rank statistics. J. Am. Stat. Assoc. (1972) 67:687–690.[CrossRef][Web of Science]
Bennett E. T. Notice of a new genus of viverridous Mammalia from Madagascar. Proc. Zool. Soc. Lond. (1833) 1833:46.
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.
de Blainville H. Des Viverras. 3. du Cryptoprocta ferox. In: Ostéographie des mammifères (1842) Paris: Baillière. Pages 15–16.
Brooks D. R. Explanations of homoplasy at different levels of biological organization. In: Homoplasy. The recurrence of similarity in evolution—Sanderson M. J., Hufford L., eds. (1996) San Diego: Academic Press. Pages 3–36.
Bugge J. The cephalic arterial system in carnivores with special referance to the systematic classification. Act. Anat. (1978) 101:45–61.[CrossRef]
Butler P. M. On the evolution of the skull and teeth in the Erinaceidae, with special reference to fossil material in the Brisith Museum. Proc. Zool. Soc. Lond. (1948) 118:446–500.
Carlsson A. Über die Systematishe Stellung von Nandinia binotata. Zool. Jahr. (1900) 13:509–528.
Carlsson A. Über Cryptoprocta ferox. Zool. Jahr. Abt. Syst. (1911) 30:419–470.
Coetzee C. G. Order Carnivora. Part 8. In: 1971–1977. The mammals of Africa: An identification manual—Meester J., Setzer H. W., eds. (1977) Washington, DC: Smithsonian Institute Press.
Corbet G. B., Hill J. E. A world list of mammalian species. (1991) London: Natural History Museum Publications and Oxford University Press.
Crawford-Cabral J. The classification of the genets (Carnivora, Viverridae, genus Genetta). Bolm Soc. port. Ciênc. nat. (1981) 20:97–114.
Crawford-Cabral J. A comment on the systematic position of Poiana. Small Carnivore Conservation (1993) 9:8.
Crouch J. E. Text-atlas of cat anatomy. (1969) Philadelphia: Lea & Febiger.
Cunningham C. W. Can three incongruence tests predict when data should be combined? Mol. Biol. Evol. (1997) 14:733–740.[Abstract]
Cunningham C. W., Omland K. E., Oakley T. H. Reconstructing ancestral character states: A critical reappraisal. T.R.E.E. (1998) 13:361–366.
D'Haese C. Is psammophily an evolutionary dead end? A phylogenetic test in the genus Willemia (Collembola; Hypogastruridae). Cladistics. (2000) 16:255–273.[CrossRef][Web of Science]
De Beaumont G. Remarques sur la classification des Felidae. Ecol. Geol. Helv. (1964) 57:837–845.
De Beaumont G. Observations sur les Herpestinae (Viverridae, Carnivora) de l'Oligocène supérieur avec quelques remarques sur des Hyaenidae du Néogène. Arch. Sci Gen. (1967) 20:79–108.
Debat V., David P. Mapping phenotypes: Canalization, plasticity and developmental stability. T.R.E.E. (2001) 16:555–561.
Desutter-Grandcolas L. Studies in cave life evolution: A rationale for future theoretical developments using phylogenetic inference. J. Zool. Syst. Evol. Res. (1997) 35:23–32.
Doyère P. L. Description d'un nouveau genre de mammifères carnassiers. Bull. Soc. Sci. Nat. (1835) 3:45.
Ewer R. F. The Carnivores. (1973) London: Wendenfeld & Nicolson.
Farris J. S. The Retention Index and homoplasy excess. Syst. Zool. (1989a) 38:406–407.
Farris J. S. The Retention Index and the Rescaled Consistency Index. Cladistics (1989b) 5:417–419.[CrossRef][Web of Science]
Flower W. H. On the value of the characters of the base of the cranium in the classification of the order Carnivora, and on the systematic position of Bassaris and other disputed forms. Proc. Zool. Soc. Lond. (1869) 1869:4–37.
Flynn J. J. Carnivoran phylogeny and rates of evolution: Morphological, taxic, and molecular. In: Carnivore behavior, ecology and evolution—Gittleman J. L., ed. (1996) Ithaca and London: Cornell University Press. Pages 542–581.
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. Nov. (1982) 2632:1–16.
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.[CrossRef][Web of Science][Medline]
Flynn J. J., Neff N. A., Tedford R. H. Phylogeny of the Carnivora. In: The phylogeny and classification of the Tetrapods—Benton M. J., ed. (1988) Oxford: Clarendon Press. Pages 73–116.
Forey P. L., Kitching I. J. Experiments in coding multistate characters. In: Homology and systematics. Coding characters for phylogenetic analysis—Scotland R. W., Pennington R. T., eds. (2000) Vol. 58. London: Taylor & Francis. Pages 54–80. Systematics Association Special.
Gatesy J., Hayashi C., Cronin M. A., Arctander P. Evidence from milk caseine genes that cetaceans are close relatives of hippopotamid artiodactyls. Mol. Biol. Evol. (1996) 13:954–963.[Abstract]
Gaubert P. Description of a new species of genet (Carnivora; Viverridae; genus Genetta) and taxonomic revision of forest forms related to the Large-spotted Genet complex. Mammalia (2003a) 67:85–108.
Gaubert P. Systématique et phylogénie du genre Genetta et des énigmatiques "genet-like taxa "Prionodon, Poiana et Osbornictis (Carnivora, Viverridae): Caractérisation de la sous-famille des Viverrinae et étude des patrons de diversification au sein du continent africain. (2003b) Paris: Muséum National d'Histoire Naturelle. Unpublished PhD Thesis.
Gaubert P., Fernandes C. A., Bruford M. W., Veron G. Genets (Carnivora, Viverridae) in Africa: An evolutionary synthesis based on cytochrome b sequences and morphological characters. Biol. J. Linn. Soc. (2004a) 81:589–610.[CrossRef][Web of Science]
Gaubert P., Tranier M., Delmas A.-S., Colyn M., Veron G. First molecular evidence for reassessing phylogenetic affinities between genets (Genetta) and the enigmatic genet-like taxa Osbornictis, PoianaPrionodon (Carnivora, Viverridae). Zool. Scr. (2004b) 33:117–129.[CrossRef][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. Biol. Sci. (2003) 270:2523–2530.[CrossRef][Medline]
Gaubert P., Veron G., Colyn M., Dunham A., Shultz S., Tranier M. A reassessment of the distributional range of the rare Genetta johnstoni (Viverridae, Carnivora), with some newly discovered specimens. Mamm. Rev. (2002b) 32:132–144.[CrossRef]
Gaubert P., Veron G., Tranier M. Genets and "genet-like" taxa (Carnivora, Viverrinae): Phylogenetic analysis, systematics and biogeographic implications. Zool. J. Linn. Soc. (2002a) 134:317–334.[CrossRef]
Ginsburg L. Plantigradie et digitigradie chez les Carnivores Fissipèdes. Mammalia (1961) 25:1–21.
Ginsburg L. Order Carnivora. In: The Miocene land Mammals of Europe—Rössner G. E., Hessig K., eds. (1999) Munich: Verlag. Pages 109–148.
Goodacre S. L., Wade C. M. Molecular evolutionary relationships between partulid land snails of the Pacific. Proc. R. Soc. Lond. B (2001) 268:1–7.[Medline]
Gorman M. L., Nedwell D. B., Smith R. M. An analysis of the anal scent pockets of Herpestes auropunctatus (Carnivora: Viverridae). J. Zool. (1974) 172:389–399.[Web of Science]
Gould S. J., Lewontin R. C. The spandrels of San Marco and the panglossian paradigm: A critique of the adaptationist programme. Proc. R. Soc. Lond. B (1979) 205:581–598.[Medline]
Graham S. W., Kohn J. R., Morton B. R., Eckenwalder J. E., Barrett S. C. H. Phylogenetic congruence and discordance among one morphological and three molecular data sets from Pontedieraceae. Syst. Biol. (1998) 47:545–567.
Grant T., Kluge A. G. Data exploration in phylogenetic inference: Scientific, heuristic, or neither. Cladistics (2003) 19:379–418.[CrossRef][Web of Science]
Gray J. E. A revision of the genera and species of Viverrine animals (Viverridae) founded on the collection in the British Museum. Proc. Zool. Soc. Lond. (1864) 1864:502–579.
Gregory W. K., Hellman M. On the evolution and major classification of the civets (Viverridae) and allied fossil and recent Carnivora: A phylogenetic study of the skull and dentition. Proc. Am. Phil. Soc. (1939) 81:309–392.
Hawkins J. A. A survey of primary homology assessment: Different workers perceive and define characters in different ways. In: Homology and systematics: Coding characters for phylogenetic analysis—Scotland R. W., Pennington R. T., eds. (2000) London: Taylor and Francis. Pages 22–53.
Hedges S. B., Sibley C. G. Molecule vs. morphology in avian evolution: The case of the "pelecaniform" birds. Proc. Natl. Acad. Sci. USA (1994) 91:9861–9865.
Hemmer H. The evolutionary systematics of living Felidae: Present status and current problems. Carnivore (1978) 1:71–79.[Web of Science]
Hillis D. M. Discriminating between phylogenetic signal and random noise in DNA sequences. In: Phylogenetic analysis of DNA sequences—Miyamoto M. M., Cracraft J., eds. (1991) Oxford: Oxford University Press. Pages 278–294.
Hillis D. M. Inferring complex phylogenies. Nature (1996) 383:130–131.[CrossRef][Medline]
Hillis D. M. Taxonomic sampling, phylogenetic accuracy, and investigator bias. Syst. Biol. (1998) 47:3–8.
Hillis D. M., Bull J. J. An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Syst. Biol. (1993) 42:182–192.
Hillis D. M., Huelsenbeck J. P. Signal, noise, and reliability in molecular phylogenetic analysis. J. Hered. (1992) 83:189–195.
Hlusko L. J. Integrating the genotype and phenotype in hominid paleontology. Proc. Natl. Acad. Sci. USA (2004) 101:2653–2657.
Hollander M., Wolfe D. A. Nonparametric statistical inference. (1973) New York: J. Wiley & Sons.
Horsfield T. Zoological researches in Java, and the neighbouring islands (1824) London: Parbury, & Allen. Printed by Kingsbury.
Hough J. R. The auditory region in some members of the Procyonidae, Canidae, and Ursidae. Bull. Am. Mus. Nat. Hist. (1948) 92:67–118.
Hunt R. M. Jr. The auditory bulla in Carnivora: An anatomical basis for reappraisal of Carnivore evolution. J. Morphol. (1974) 143:21–76.[CrossRef][Web of Science][Medline]
Hunt R. M. Jr. Evolution of the Aeluroid Carnivora: Significance of auditory structure in the Nimravid Cat Dinictis. Am. Mus. Nov. (1987) 2886:1–74.
Hunt R. M. Jr. Evolution of the Aeluroid Carnivora: Significance of the ventral promontorial process of the petrosal, and the origin of basicranial patterns in the living families. Am. Mus. Nov. (1989) 2930:1–32.
Hunt R. M. Jr. Evolution of the aeluroid Carnivora: Viverrid affinities of the Miocene carnivoran Herpestides. Am. Mus. Nov. (1991) 3023:1–34.
Hunt R. M. Jr. Biogeography of the Order Carnivora. In: Carnivore behavior, ecology and evolution—Gittleman J. L., ed. (1996) New York: Cornell University Press. Pages 451–485.
Hunt R. M. Jr. Evolution of the Aeluroid Carnivora: Diversity of the earliest Aeluroids from Eurasia (Quercy, Hsanda-Gol) and the origin of Felids. Am. Mus. Nov. (1998) 3252:1–65.
Hunt R. M. Jr. Basicranial anatomy of the living Linsangs PrionodonPoiana (Mammalia, Carnivora, Viverridae), with comments on the early evolution of Aeluroid Carnivora. Am. Mus. Nov. (2001) 3330:1–24.[CrossRef]
Hunt R. M. Jr., Tedford R. H. Phylogenetic relationships within the Aeluroid Carnivora and implications of their temporal and geographic distribution. In: Mammal phylogeny: Placentals—Szalay F. S., Novacek M. J., McKenna M. C., eds. (1993) New York: Springer-Verlag. Pages 53–73.
Ihaka R., Gentleman R. R: A language for data analysis and graphics. J. Comp. Graph. Stat. (1996) 5:299–314.[CrossRef]
Ivanoff D. Partitions in the carnivoran auditory bulla: Their formation and significance for systematics. Mamm. Rev. (2001) 31:1–16.[CrossRef]
Jernvall J., Keränen S. V. E., Thesleff I. Evolutionary modification of development in mammalian teeth: Quantifying gene expression patterns and topography. Proc. Natl. Acad. Sci. USA (2000) 97:14444–14448.
King J. E. Relationships of the hooded and elephant seals (genera CystophoraMirounga). J. Zool. (1966) 148:385–398.[Web of Science]
Kingdon J. East african Mammals: An atlas of evolution in Africa. (1977) London: Academic Press.
Kingdon J. The Kingdon field guide to African mammals (1997) San Diego: Academic Press.
Kleiman D. G. Scent marking in the binturong, Arctictis binturong. J. Mamm. (1974) 55:224–227.[CrossRef][Web of Science]
Kluge A. G. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. (1989) 38:7–25.[Abstract]
Kluge A. G. Testability and the refutation and corroboration of cladistic hypotheses. Cladistics (1997) 13:81–96.[CrossRef][Web of Science]
Kusmierski R., Borgia G., Uy A., Crozier R. H. Labile evolution of display traits in bowerbirds indicates reduced effects of phylogenetic constraint. Proc. R. Soc. Lond. B (1997) 264:307–313.[CrossRef][Medline]
Laborde C. Caractères d'adaptation des membres au mode de vie arboricole chez Cryptoprocta ferox par comparaison avec d'autres Carnivores Viverridés. Ann. Sci. Nat. Zool. Paris (1987) 8:25–39.
Leal M., Know A. K., Losos J. B. Lack of convergence in aquatic Anolis lizard. Evolution. (2002) 56:785–791.[Web of Science][Medline]
Lee M. S. Y. Uninformative characters and apparent conflict between molecules and morphology. Mol. Biol. Evol. (2001) 18:676–680.
Lee M. S. Y., Hugall A. L. Partitioned likelihood support and the evaluation of data set conflict. Syst. Biol. (2003) 52:15–22.
Lewis P. Phylogenetic systematics turns over a new leaf. T.R.E.E. (2001) 16:30–37.[Medline]
Li W.-H. Molecular evolution (1997) Sunderland, Massachusetts: Sinauer Associates.
Littlewood D. T. J., Rohde K., Bray R. A., Herniou E. A. Phylogeny of the Platyhelminthes and the evolution of parasitism. Biol. J. Linn. Soc. (1999) 68:257–287.[CrossRef][Web of Science]
Lovejoy C. O., Cohn M. J., White T. D. Morphological analysis of the mammalian postcranium: A devlopmental perspective. Proc. Natl. Acad. Sci. USA (1999) 96:13247–13252.
Macey J. R., Schulte J. A. II, Larson A., Tuniyev B. S., Orlov N., Papenfuss T. J. Molecular phylogenetics, tRNA evolution, and historical biogeography in Anguid lizards and related taxonomic families. Mol. Phylogenet. Evol. (1999) 12:250–272.[CrossRef][Web of Science][Medline]
Maddison W. P., Maddison D. R. MacClade: Analysis of phylogeny and character evolution, version 3. (1992) Sunderland, Massachusetts: Sinauer Associates.
Maddison W. P., Maddison D. R. MacClade: Analysis of phylogeny and character evolution, version 4.0. (2000) Sunderland, Massachusetts: Sinauer Associates.
Mardulyn P., Whitfield J. B. Phylogenetic signal in the COI, 16S, and 28S genes for inferring relationships among genera of Microgastrinae (Hymenoptera; Braconidae): Evidence of a high diversification rate in this group of parasitoids. Mol. Phylogenet. Evol. (1999) 12:282–294.[CrossRef][Web of Science][Medline]
Mattern M. Y., McLennan D. A. Phylogeny and speciation of Felids. Cladistics (2000) 16:232–253.[CrossRef][Web of Science]
McKenna M. C., Bell S. K. Classification of mammals above the species level. (1997) New York: Columbia University Press.
McCracken K. G., Harshman J., McClellan D. A., Afton A. D. Data set incongruence and correlated character evolution: An example of functional convergence in the hind-limbs of stifftail diving ducks. Syst. Biol. (1999) 48:683–714.
Mivart St.-G. On the classification and distribution of the Ailuroidea. Proc. Zool. Soc. Lond. (1882) 1882:135–208.
Miyamoto M. M., Fitch W. M. Testing species phylogenies and phylogenetic methods with congruence. Syst. Biol. (1995) 44:64–76.
Neff N. A. The basicranial anatomy of the Nimravidae (Mammalia, Carnivora): Character analyses and phylogenetic inferences. (1983) New York: The City University of New York. Unpublished PhD Thesis.
Novacek M. J. Aspects of the problem of variation, origin and evolution of the eutherian auditory bulla. Mamm. Rev. (1977) 7:131–150.[CrossRef]
Novacek M. J. The skull of leptictid insectivorans and the higher-level classification of eutherian mammals. Bull. Amer. Mus. Nat. Hist. (1986) 183:1–112.
Nowak R. M. Walker's mammals of the world (1999) 6th ed. Baltimore: The Johns Hopkins University Press.
Parra-Olea G., Wake D. B. Extreme morphological and ecological homoplasy in tropical salamanders. Proc. Natl. Acad. Sci. USA (2001) 98:7888–7891.
Patterson C., Johnson G. D. The data, the matrix, and themessage: Comment on Begle's "Relationships of the osmeroid fishes." Syst. Biol. (1997) 46:358–365.
Peigné S., De Bonis L. Cranial anatomy of Nimravidae (Mammalia, Carnivora): Biological and phylogenetic implications. Zool. J. Linn. Soc. (2003) 138:477–493.[CrossRef]
Petter G. Interprétation évolutive de la denture des Viverridés africains. Mammalia (1969) 33:607–625.
Petter G. Rapports phylétiques des Viverridés (Carnivores, Fissipèdes). Les formes de Madagascar. Mammalia (1974) 38:605–636.
Pimental R. A., Riggins R. The nature of cladistic data. Cladistics (1987) 3:275–289.
Pleijel F. On character coding for phylogenetic reconstruction. Cladistics (1995) 11:309–315.[CrossRef][Web of Science]
Pocock R. I. On the feet and glands and other external characters of the Paradoxurinae Genera ParadoxurusArctictisArctogalidiaNandinia. Proc. Zool. Soc. Lond. (1915a) 1915:387–412.
Pocock R. I. On some of the external characters of Cynogale bennettii. Gray. Proc. Zool. Soc. Lond. (1915b) 1915:350–360.
Pocock R. I. On some of the external characters of the genus Linsang, with notes upon the genera Poiana and Eupleres. Ann. Mag. Nat. Hist. (1915c) 16(8):341–351.
Pocock R. I. On some of the external characters of the palm-civet (Hemigalus derbyanus, Gray) and its allies. Ann. Mag. Nat. Hist. (1915d) 16(8):153–162.
Pocock R. I. On the feet and glands and other external characters of the Viverrinae, with the description of a new genus. Proc. Zool. Soc. Lond. (1915e) 1915:131–149.
Pocock R. I. On some external characters of Galidia, Galidictis, and related genera. Ann. Mag. Nat. Hist. (1915f) 16(8):351–356.
Pocock R. I. On the external characters of the mongooses (Mungotidae). Proc. Zool. Soc. Lond. (1916a) 1:349–374.
Pocock R. I. On some of the external characters of Cryptoprocta. Ann. Mag. Nat. Hist. (1916b) 17(8):413–425.
Pocock R. I. On some of the external structural characters of the striped hyaena (Hyaena hyaena) and related genera and species. Proc. Zool. Soc. Lond. (1916c) 1916:330–343.
Pocock R. I. On the external characters of the mongooses (Mungotidae). Proc. Zool. Soc. Lond. (1916d) 1:349–374.
Pocock R. I. The rarer genera of Oriental Viverridae. Proc. Zool. Soc. Lond. (1933) 1933:969–1035.
Poe S., Wiens J. J. Character selection and the methodology of morphological phylogenetics. In: Phylogenetic analysis of morphological data—Wiens J. J., ed. (2000) Washington, DC: Smithsonian Institution Press. Pages 20–36.
Prychitko T. M., Moore W. S. Comparative evolution of the mitochondrial cytochrome b gene and nuclear B-fibrinogen intron 7 in woodpeckers. Mol. Biol. Evol. (2000) 17:1101–1111.
Purvis A., Quicke D. L. J. Phylogeny reconstruction: Are the big easy? T.R.E.E. (1997) 12:49–50.
Qiu Z.-X. Die Hyaeniden aus dem Ruscinium und Villafranchium Chinas. Müncher Geowissenschaftliche Abhandlungen Reihe A. Geol. Palänotol. (1987) 9:1–110.
Quicke D. L. J., Belshaw R. Incongruence between morphological data sets: An example from the evolution of endoparasitism among parasitic wasps (Hymenoptera: Braconidae). Syst. Biol. (1999) 48:436–454.
Radinsky L. B. An example of parallelism in carnivore brain evolution. Evolution (1971) 25:518–522.[CrossRef][Web of Science]
Radinsky L. B. Viverrid neuroanatomy: Phylogenetic and behavioral implications. J. Mamm. (1975) 56:130–150.[CrossRef][Web of Science][Medline]
Robinette H. R., Stains H. J. Comparative study of the calcanea of the Pinnipedia. J. Mamm. (1970) 51:527–541.[CrossRef][Web of Science]
Roeder J.-J. Marking behaviour and olfactory recognition in genets (Genetta genetta L. Carnivora-Viverridae). Behaviour. (1980) 72:200–210.[CrossRef]
Roeder J.-J. Contribution à l'étude des modalités de communication chez la Genette. (1984) Strasbourg, France: Université Louis Pasteur. Unpublished PhD Thesis.
Rosevear D. R. The Carnivores of West Africa. (1974) London: Trustees of the British Museum (Natural History.
Royston P. An extension of Shapiro and Wilk's W test for normality to large samples. Appl. Stat. (1982a) 31:115–124.[CrossRef]
Royston P. Algorithm AS 181: The W test for normality. Appl. Stat. (1982b) 31:176–180.[CrossRef]
Royston P. A remark on algorithm AS 181: The W test for normality. Appl. Stat. (1995) 44:547–551.[CrossRef]
Ruta M., Jeffery J. E., Coates M. I. A supertree pf early tetrapods. Proc. R. Soc. Lond. B (2003) 270:2507–2516.[CrossRef][Medline]
Salazar-Ciudad I., Jernvall J. A gene network model accounting for development and evolution of mammalian teeth. Proc. Natl. Acad. Sci. USA (2002) 99:8116–8120.
Savolainen V. S., Chase M. W., Salamin N., Soltis D. E., Soltis P. S., Lopez A. J., Fedrigo O., Naylor G. J. P. Phylogeny reconstruction and functional constraints in organellar genomes: Plastid atpBrbcL sequences versus animal mitochondrion. Syst. Biol. (2002) 51:638–647.
Scotland R. W., Olmstead R. G., Bennett J. R. Phylogeny reconstruction: The role of morphology. Syst. Biol. (2003) 52:539–548.
Sechrest W., Brooks T. M., da Fonseca G. A. B., Konstant W. R., Mittermeier R. A., Purvis A., Rylands A. B., Gittleman J. L. Hotspots and the conservation of evolutionary history. Proc. Natl. Acad. Sci. USA (2002) 99:2067–2071.
Seravin L. N. The principle of counter-directional morphological evolution and its significance for construction the megasystem of protists and other eukaryotes. Protistology (2001) 2:6–14.
Shoshani J. On the phylogenetic relationships among Paenungulata and within Elephantidae as demonstrated by molecular and osteological evidence. (1986) Detroit, Michigan: Wayne State University. Unpublished PhD Thesis.
Simpson G. G. The principles of classification and a classification of Mammals. Bull. Am. Mus. Nat. Hist. (1945) 85:1–350.
Smith A. B. Systematics and the fossil record: Documenting evolutionary patterns. (1994) Oxford, U.K.: Blackwell Scientific.
Sorenson M. D. TreeRot, version 2. (1999) Boston: Boston University.
Stevens P. F. Character states, morphological variation, and phylogenetic analysis: A review. Syst. Bot. (1991) 16:553–583.[CrossRef]
Swofford D. L. PAUP*: Phylogenetic analysis using parsimony (* and other methods). Beta version 4.0b2. (2001) Sunderland, Massachusetts: Sinauer Associates.
Swofford D. L., Maddison W. P. Reconstructing ancestral character states under Wagner parsimony. Math. Biosci. (1987) 87:199–229.[CrossRef][Web of Science]
Taylor M. E. The functional anatomy of the forelimb of some African Viverridae (Carnivora). J. Morphol. (1974) 143:307–333.[CrossRef][Web of Science][Medline]
Taylor M. E. The functional anatomy of the hindlimb of some African Viverridae (Carnivora). J. Morphol. (1976) 148:227–253.[CrossRef][Web of Science][Medline]
Taylor M. E. Foot structure and phylogeny in the Viverridae (Carnivora). J. Zool. (1988) 216:131–139.[CrossRef][Web of Science]
Teeling E. C., Madsen O., Van Den Bussche R. A., de Jong W. W., Stanhope M. J., Springer M. S. Microbat paraphyly and the convergent evolution of a key innovation in Old World rhinolophoid microbats. Proc. Natl. Acad. Sci. USA (2002) 99:1431–1436.
Teilhard d e, Chardin P. Les Carnassiers des Phosphorites du Quercy. Ann. Pal. (1915) 9:103–191.
Templeton A. R. Phylogenetic inference from restriction site endonuclease cleavage site maps with particular reference to the humans and apes. Evolution (1983) 37:221–244.[CrossRef][Web of Science]
Van Valkenburg B. Skeletal indicators of locomotor bahavior in living and extinct carnivores. J. Vert. Pal. (1987) 7:162–182.
Veron G. Méthodes de recherche en biotaxinomie des mammifères carnivores—Confrontation des méthodes de phylogénie traditionnelle et moléculaire dans la recherche de la position systématique de Cryptoprocta ferox (Aeluroidea). (1994) Paris: Muséum National d'Histoire Naturelle. Unpublished PhD Thesis.
Veron G. La position systématique de Cryptoprocta ferox (Carnivora). Analyse cladistique des caractères morphologiques de carnivores Aeluroidea actuels et fossiles. Mammalia (1995) 59:551–582.
Veron G. Pads morphology in the Viverridae (Carnivora). Act. Ther. (1999) 44:363–376.
Veron G., Catzeflis F. Phylogenetic relationships of the endemic Malagasy Carnivore Cryptoprocta ferox (Aeluroidea): DNA/DNA hybridization experiments. J. Mamm. Evol. (1993) 1:169–185.[CrossRef]
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]
Veron G., Gaubert P. The Mammals of Africa. In: Carnivora, Pinnipedia, Pholidota, Tubulidentata, Hyracoidea, Proboscidea, Sirenia, Perissodactyla—Kingdon J., Butynski T., eds. Vol. 4. Oxford: Elsevier Science. in press.
Veron G., Heard S. Molecular systematics of the Asiatic Viverridae (Carnivora) inferred from mitochondrial cytochrome b sequence analysis. J. Zool. Syst. Evol. Res. (2000) 38:209–217.[CrossRef]
Wagner G. P. The character concept in evolutionary biology. (2001) San Diego: Academic Press.
Wayne R. K., Benveniste R. E., Janczewski D. N., O'Brien S. J. Molecular and biochemical evolution of the Carnivora. In: Carnivore behavior, ecology and evolution.—Gittleman J. L., ed. (1989) New York and London: Cornell University Press. Pages 465–498.
Wemmer C. M. Comparative ethology of the large–spotted genet Genetta tigrina and some related Viverrids. (1977) Washington, DC: Smithsonian Institution Press.
Wenzel J. W., Siddall M. E. Noise. Cladistics (1999) 15:51–64.[CrossRef][Web of Science]
Werdelin L. Carnivoran ecomorphology: A phylogenic perspective. In: Carnivore behavior, ecology and evolution—Gittleman J. L., ed. (1996) Vol. 2. Ithaca, New York: Cornell University Press. Pages 583–624.
Whitlock B. A., Baum D. A. Phylogenetic relationships of Theobroma and Herrania (Sterculiaceae) based on sequences of the nuclear gene Vicilin. Syst. Bot. (1999) 24:128–138.[CrossRef]
Wiens J. Combining data sets with different phylogenetic histories. Syst. Biol. (1998) 47:568–581.
Wiens J. J., Chippindale P. T., Hillis D. M. When are phylogenetic analyses misled by convergence? A case study in Texas cave salamanders. Syst. Biol. (2003) 52:501–514.
Wilkinson M. A comparison of two methods of character construction. Cladistics (1995) 11:297–308.[CrossRef][Web of Science]
Wilson E. O. Sociobiology. In: The new synthesis. (1975) Cambridge, Massachusetts: Harvard University Press.
Windle B. C. A., Parsons F. G. On the mycology of the terrestrial Carnivora. Proc. Zool. Soc. Lond. (1897) 1897:370–409.
Wozencraft W. C. A phylogenetic reappraisal of the Viverridae and its relationship to other Carnivora. (1984) Lawrence: Univirsity of Kansas. Unpublished PhD Thesis.
Wozencraft W. C. Classification of the recent Carnivora. In: Carnivore behavior, ecology, and evolution.—Gittleman J. L., ed. (1989) Ithaca, New York: Cornell University Press. Pages 569–593.
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 and London, DC: Smithsonian Institution Press. Pages 279–348.
Wurster D. H. Cytogenetic and phylogenetic studies in Carnivore. In: Comparative mammalian cytogenetics—Benirschke K., ed. (1969) Berlin: Springer-Verlag. Pages 310–329.
Wyss A. R. The walrus auditory region and the monophyly of pinnipeds. Am. Mus. Nov. (1987) 2871:1–31.
Wyss A. R., Flynn J. J. A phylogenetic analysis and definition of the Carnivora. In: Mammal phylogeny: Placentals—Szalay F. S., Novacek M. J., McKenna M. C., 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]
Zakharov V. M. Population phenogenetics: Analysis of developmental stability in natural populations. Acta Zool. Fenn. (1992) 191:7–30.
Zar J. H. Biostatistical analysis. (1984) Englewood Cliffs, New Jersey: Prentice-Hall.
Zaragüeta i, Bagils R., Gallut C., Vignes-Lebbe R. Noisesnapper: A method for detecting noise in phylogenetic analysis. 21st Annual Meeting of the Willi Hennig Society, Cladistics 19. Page 162.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. L. J. Quicke, O. R. Jones, and D. R. Epstein Correcting the Problem of False Incongruence Due to Noise Imbalance in the Incongruence Length Difference (ILD) Test Syst Biol, June 1, 2007; 56(3): 496 - 503. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. H. Wortley and R. W. Scotland The Effect of Combining Molecular and Morphological Data in Published Phylogenetic Analyses Syst Biol, August 1, 2006; 55(4): 677 - 685. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||