The systematics of the Cervidae: a total evidence approach

Abstract

Systematic relationships of cervids have been controversial for decades. Despite new input from molecular systematics, consensus could only be partially reached. The initial, gross (sub) classification based on morphology and comparative anatomy was mostly supported by molecular data. The rich fossil record of cervids has never been extensively tested in phylogenetic frameworks concerning potential systematic relationships of fossil cervids to extant cervids. The aim of this work was to investigate the systematic relationships of extant and fossil cervids using molecular and morphological characters and make implications about their evolutionary history based on the phylogenetic reconstructions. To achieve these objectives, molecular data were compiled consisting of five nuclear markers and the complete mitochondrial genome of 50 extant and one fossil cervids. Several analyses using different data partitions, taxon sampling, partitioning schemes, and optimality criteria were undertaken. In addition, the most extensive morphological character matrix for such a broad cervid taxon sampling was compiled including 168 cranial and dental characters of 41 extant and 29 fossil cervids. The morphological and molecular data were analysed in a combined approach and other comprehensive phylogenetic reconstructions. The results showed that most Miocene cervids were more closely related to each other than to any other cervids. They were often positioned between the outgroup and all other cervids or as the sister taxon to Muntiacini. Two Miocene cervids were frequently placed within Muntiacini. Plio- and Pleistocene cervids could often be affiliated to Cervini, Odocoileini or Capreolini. The phylogenetic analyses provide new insights into the evolutionary history of cervids. Several fossil cervids could be successfully related to living representatives, confirming previously assumed affiliations based on comparative morphology and introducing new hypotheses. New systematic relationships were observed, some uncertainties persisted and resolving systematics within certain taxa remained challenging.

Keywords: Combined analyses; Evolutionary history; Morphology; Phylogeny; Ruminantia.

Figure 1. Diagnostic cranial characters of cervids.

The most important diagnostic cranial features of cervids, with the exception of antlers, which almost exclusively occur in males, are outlined in this figure as (A) a photograph and (B) a drawing of the cranium of a female Blastocerus dichotomus (MNHN 1933-207). Note the brachyodont dentition, the preorbital vacuity, lacrimal fossa and lacrimal foramina. (Drawing by Nicola Heckeberg) pmx, premaxillary; mx, maxillary; nas, nasal; lac, lacrimal; zyg, zygomaticum; pal, palatine; pte, pterygoid; orb, orbitosphenoid; fro, frontal; par, parietal; ali, alisphenoid; squ, squamosal; soc, supraoccipital; ppa, paroccipital processes; bul, auditory bulla; con, condyles.

Figure 2. Age ranges of fossil cervids.

Fossil cervids are arranged from the youngest first appearance datum to the oldest first appearance datum (left/top). The stage column widths are not to scale with time. The dates were compiled from the literature (Gentry, Rössner & Heizmann, 1999; Steininger, 1999; Böhme et al., 2012; Hilgen et al., 2012; Cohen et al., 2013; Croitor, 2014) and databases (NOW: www.helsinki/science/now/, PBDB: www.paleobiodb.org).

Figure 3. Overview of all analyses.

This overview shows all analyses undertaken and the optimality criteria under which they were run. BI, Bayesian inference; ML, maximum likelihood; MP, maximum parsimony.

Figure 4. Overview of the characters available for each cervid species.

Extant species are arranged in alphabetical order; fossil cervids and the two non-cervid fossils are arranged from the youngest to the oldest following the extant taxa. Morphological characters are subdivided into seven partitions indicated by the different colours of each bar. The y-axis represents the absolute number of present characters.

Figure 5. Details of the cervid cranial anatomy.

(A) Cranium of Dicrocerus elegans (MNHN Sa 10308) in dorsal view. The arrow indicates the sagittal crest. (B) Basicranium of Odocoileus hemionus (MNHN AE724). The arrow indicates the vomerine septum typical for Capreolinae. (C) Basicranium of Axis axis (ZSM 1958-88). The arrows indicate the large inflated auditory bullae, rarely observed in cervids. (D) Basicranium of Ozotoceros bezoarticus (UMZC H.18781). The arrows indicate the small flattened auditory bullae with prominent processes.

Figure 6. Cervid antler evolution.

(A) Cranium of a typical Miocene cervid showing the characteristics of early pedicles and antlers. (B) Cranium of an extant cervids showing features of the pedicles and antlers seen in modern cervids (Drawing by Nicola Heckeberg).

Figure 7. Details of the upper dentition.

Close ups of the upper dentition of selected cervids showing the most striking features. (A) Rucervus duvaucelii (ZSM 1957-60), (B) Rangifer tarandus (ZSM 1959-211), (C) Rucervus eldii (UMZC H16194), (D) Elaphurus davidianus (UMZC H16235) and (E) Odocoileus hemionus (ZSM 1971-720).

Figure 8. Details of the lower dentition.

Close ups of the lower dentition of selected cervids showing the most striking features. (A) Rucervus duvaucelii (ZSM 1957-60), (B) Rangifer tarandus (ZSM 1959-211), (C) Rucervus eldii (UMZC H16194), (D) ‘Cervus’ philisi (NMB St.V. 605) and (E) Procervulus (MNHN LRM 114).

Figure 9. Variability of p4 in cervids.

This sequence of the lower left p4 shows different degrees of modification of tooth elements, starting with an open anterior valley (A), development of mesolingual cristids (B), connexion of mesolingual cristids to other tooth elements (C), closing of the anterior valley (D) and re-arrangement of tooth elements with a diagonal orientation (E). The premolars with modifications shown in (D) and (E) resemble the molars (Drawing by Nicola Heckeberg).

Figure 10. Colour code.

The colour code provides the key to taxonomic groups studied here and applies to all topologies within the present work.

Figure 11. Topologies from the morphological analyses.

The topologies of the maximum parsimony analyses of the combined morphological data set are shown. (A) Topology based on the unordered data set, (B) topology based on the analysis using character state ordering. Node support values are given as bootstrap support values.

Figure 12. Topologies from the SFA and FPA approaches.

(A) The topology summarises the systematic position of the fossils based on the SFA (see Supplemental File S3 for individual topological placements). The gradually shaded boxes indicate different observed positions, e.g. Procapreolus cusanus was placed within Capreolinae in one analysis and within Odocoileini in a different analysis. (B) The topology is the result of the FPA analysis.

Figure 13. Topologies resulting from the molecular data sets.

The topologies of the Bayesian inference analyses of the combined nuclear data set (A), the combined mitochondrial data set (B) and the combined molecular data set (C) are shown. Nuclear markers were available for fewer taxa than mitochondrial markers. The mitochondrial and molecular combined topologies are congruent except for the position of Alces alces and Pudu mephistophiles. Node support values are given as Bayesian posterior probabilities.

Figure 14. Topologies from the combined molecular and morphological analyses.

The topologies of the combined molecular and morphological analyses using Bayesian inference (A), maximum likelihood (B) and maximum parsimony (C) are shown. Node support values are Bayesian posterior probabilities (BI) and bootstrap support values (ML and MP).

Figure 15. The qualitative summary topology of all analyses is shown.

(A) Represents the overview of the systematic relationships of higher cervid taxa including the positions of some fossil cervids, (B) shows the systematic relationship of several Plio- and Pleistocene cervids and (C) shows the systematic relationships of Miocene cervids.