Baleen boom and bust: a synthesis of mysticete phylogeny, diversity and disparity

A new, fully dated total-evidence phylogeny of baleen whales (Mysticeti) shows that evolutionary phases correlate strongly with Caenozoic modernization of the oceans and climates, implying a major role for bottom-up physical drivers. The phylogeny of 90 modern and dated fossil species suggests three major phases in baleen whale history: an early adaptive radiation (36–30 Ma), a shift towards bulk filter-feeding (30–23 Ma) and a climate-driven diversity loss around 3 Ma. Evolutionary rates and disparity were high following the origin of mysticetes around 38 Ma, coincident with global cooling, abrupt Southern Ocean eutrophication and the development of the Antarctic Circumpolar Current (ACC). Subsequently, evolutionary rates and disparity fell, becoming nearly constant after approximately 23 Ma as the ACC reached its full strength. By contrast, species diversity rose until 15 Ma and then remained stable, before dropping sharply with the onset of Northern Hemisphere glaciation. This decline coincided with the final establishment of modern mysticete gigantism and may be linked to glacially driven variability in the distribution of shallow habitats or an increased need for long-distance migration related to iron-mediated changes in glacial marine productivity.


Supplementary Material
Marx FG, Fordyce, RE. Baleen boom and busta synthesis of mysticete phylogeny, diversity and disparity. Royal Society Open Science Contents Figs. S1 to S6 Table S1   Table S2 Review of stratigraphic ranges

Institutional abbreviations
List of studied material Supplementary references

Morphological characters
Note: The morphological data matrix and all images associated with this paper are available from MorphoBank (www.morphobank.org), project 687. A Nexus file containing the full supermatrix and MrBayes codes for carrying out the total evidence dating analysis are also available, and stored in the "Documents" section of the same project.       Table S1. Stratigraphic ranges of included fossil taxa. Where two ranges are shown, those in parentheses were used for the total evidence dating analysis (see Methods and review of stratigraphic ranges below). Specimen: SMAC 1309 [78,79].
Comments: The Chippubetsu Formation, which forms part of the Fukagawa Group exposed in Hokkaido, Japan [80]. The specimen was found in a horizon below a tuff level ("T2") likely correlative with the S1 tuff of the Takikawa Formation [81], which has been dated to 4.1 ± 0.6 Ma based on fission-track dating [82]. The age of the base of the Chippubetsu Formation is currently unknown. However, the lower part of this formation has been correlated with the upper part of the Horokaoshirarika Formation based on palynological data [80], with the latter falling into the Neodenticula kamtschatica diatom zone [83], dated to ca. 6.4-3.5 Ma [84]. Together, these estimates imply an approximate age range of 6.4-4.1 Ma.
Comments: The Chersonian Formation has been dated to the late Sarmatian sensu lato, based on the occurrence of numerous Mactra (Chersonimactra) caspia [85]. In the region of the Central Paratethys, the Sarmatian Stage (sensu stricto) as a whole has been assigned the Middle Miocene [86,87]. By contrast, the upper part of the Sarmatian as found in the area of the eastern Paratethys correlates with the early Tortonian (11.2-9.6 Ma) [88,89]. Mary"s Formation are likely referable to C. megalophysum [23], and indicate an almost exclusively early Tortonian (11.7-10.0 Ma) age for this taxon [48,49]. See Cephalotropis coronatus for further details.
Comments: In the region of the Central Paratethys, the Sarmatian Stage (sensu stricto) as a whole has been assigned the Middle Miocene [86,87]. By contrast, the upper part of the Sarmatian as found in the area of the eastern Paratethys correlates with the early Tortonian (11.2-9.6 Ma) [88,89]. The age of the late Sarmatian deposits on the Taman Peninsula are still a matter of debate. Fission-track dating places their lower boundary at ca. 11.2 ± 0.7 Ma, while chronostratigraphy suggest an upper limit of roughly 9.6 Ma (or possibly somewhat younger, based on an estimate of 9.5 ± 0.9 Ma for the upper Sarmatian deposits of the nearby Kerch Peninsula) [88]. By contrast, biostratigraphic data suggest an age range of 8.9 Ma to 7.9-7.7 Ma [89]. Until further data are available to resolve this conflict, we assume the age of C. rathkii to be 11.2-9.6 Ma, in line with the radiometric evidence.
Locality and horizon: Outskirts of Nikolaev, Ukraine: possibly Chersonian Formation [100]. and together with latter forms the middle/upper portion of the Bihoku Group [114].
Calcareous nannoplankton correlates the Korematsu Formation and the lower part of the Itabashi Formation with the upper part of nannoplankton zone NN4 [114], suggesting a latest Burdigalian or early Langhian age with a minimum of 14.9 Ma [20,54]. A further study [115] specifically correlated the Korematsu and Itabashi formations exposed along the stretch of the Saijyo River that yielded the whale fossils with the top of nannoplankton zone NN4, and the beginning of the "Mid-Neogene Climatic Optimum" (ca. 116]
Stratigraphic range: early? Chattian (28.1-25.6 Ma; preliminary assessment) Comments: The Jan Juc Marl is placed mostly into the Chattian by both biostratigraphy and radiometric dating, with a minimum age of 23.9 (based on the overlying Puebla Formation) and a maximum age exceeding 27.2 Ma [138]. However, the exact position of the known material within the Jan Juc Marl is unknown. Several recent cladistic analyses, including the present study, have proposed a close relationship of Janjucetus and Mammalodon [102,139]. Since the two taxa must have separated prior to the first occurrence of Mammalodon and its relatives during the early Chattian (see discussion of OU 22026 for further details), we here assume an early Chattian (28.1-25.6 Ma) age for Janjucetus, pending the discovery of temporally better constrained specimens.
Comments: The Haraichi Formation is bracketed by the underlying Baba and the overlying Kamikoizawa tuffs [140]. 40 Ar/ 39 Ar dating places the Baba Tuff at 11.3 Ma [141]. The Kamikoizawa Tuff has not yet been dated directly, but an earliest late Tortonian age (less than 11.0 Ma) has been suggested on the basis of sedimentation rates [142].
Comments: Mollusc biostratigraphy indicates the lower part of the Blinovo Formation to fall into the Middle Sarmatian [144], which in the region of the Eastern Paratethys spans the Serravallian/early Tortonian boundary, and ranges from ca. 12.1-11.2 Ma [88].
Locality and horizon: Near Coquimbo, Chile: formation not stated, but likely to be Coquimbo Formation (see comments) [155].
Comments: In his original description, Dathe [155] did not specify the formation that has yielded the specimen, but stated that the fossil had been found at the western end of the "Bahia de Guayacan", southwest of Coquimbo. He furthermore speculated that the fossil had Formation [156,157]. The latter is divided into 16 lithostratigraphic units, 5 of which (units 3, 4 and 11-13) have yielded cetaceans remains [156,158]. Of the latter, only units 4 and 11-13 contain sandstone, with only unit 11 being dominated by this lithology. Strontium dating and biostratigraphic evidence place units 3 and 4 in the Tortonian (11.9-11.2 Ma), and thus make them much older than the Early Pliocene age suggested by Dathe [155]. By contrast,  [156]. Considering the coincidence of (i) the occurrence of numerous cetacean fossils in units of the correct lithology and age; and (ii) the broad exposure of these units in the type area of "M. hubachi", we suggest that the holotype of this species was recovered from the upper, Early Pliocene portion of the Coquimbo Formation.
As a result, we follow Dathe [155] in assigning an Early Pliocene (Zanclean, 5.3-3. 6  Pending further study, these specimens may extend the range of the lineage by up to 5 Ma.
Locality and horizon: Near the mouth of the Potomac River: the formation has not been specified, but may be the St Mary"s Formation (see comments) [90] Stratigraphic range: early Tortonian (11.7-10.0 Ma).
Comments: USNM 8518 was collected from "a Miocene marl from near the mouth of the Potomac river" [90: 143]. Subsequent authors interpreted this description of the type locality to refer to either the Calvert Formation [93,163]  Member. The latter has been strontium dated to the early Tortonian (11.7-10.0 Ma) [48,49], which is the age assumed here pending the emergence of better-constrained occurrence data.
Comments: Originally referred to as a "prosqualodontid" [183], the specimen was later reassigned to Llanocetus sp. [147]. Based on its relative position within the Ototara Limestone, the horizon that yielded the specimen likely correlates with the early Whaingaroan, ca. 33.0-32.0 Ma [184].
Comments: The Pińczów Formation has been assigned to the early-middle Badenian based on foraminiferal data [195,196], and is constrained by the occurrence of Orbulina suturalis (first appearing at 15.1 Ma [20]) and its correlation with the Lagenidae benthic foraminiferal zone [197], which terminates at or near the Langhian-Serravallian boundary [198], i.e. 13 [74,75,167]), resulting in a total age range of 7.5-5.9 Ma.
Locality and horizon: Western portion of a bend in the Stirone River, approximately 2 km northwest of Salsomaggiore Terme, Italy: formation not stated [200].
Comments: The holotype was recovered from grey, clayey marls bearing invertebrate fossils indicative of a Tortonian age [200]. According to Artoni et al. [201: fig. 2], the rocks exposed along this section of the river where the specimen was found form part of a middle Eoceneearly Tortonian Epiligurian succession incorporated into an "Intra-Messinian Chaotic Unit".
Pending the publication of further data on the exact age of these strata, P. quarantellii is therefore here assumed to be early Tortonian (11.6-9.44 Ma).
Locality and horizon: Gram Clay Pit, Gram, South Jutland, Denmark: uppermost portion of the Gram Formation [207].

List of studied material
In the following list, a species or specimen marked with an asterisk was scored for the phylogenetic analysis but not illustrated on MorphoBank owing to its ongoing description or re-description by other workers.

Outgroup:
Archaeodelphis patrius Allen, 1921 47. Width of supraorbital process as measured in a straight line from the lateralmost point of the postorbital process to the intertemporal constriction: equal to or shorter than the anteroposterior length of the supraorbital process above the orbit (0); up to twice the length above the orbit (1); more than twice the length above the orbit (2). ORDERED 48. Postorbital process in dorsal view: oriented posteriorly (0); oriented laterally (1); oriented posterolaterally (2); short and not markedly projecting in any direction (3). 80. Area enclosed by orbitotemporal crest on supraorbital process of frontal: forms less than half of the dorsal surface of the supraorbital process (0); covers half or more of the dorsal surface of the supraorbital process (1). 81. Outline and orientation of orbitotemporal crest: subparallel to posterior border of supraorbital process (0); distal half oriented distinctly posterolaterally and approaching the posterolateral corner of the supraorbital process (1); as state 1, but with the crest terminating halfway along the posterior border of the supraorbital process (2); as state 2, but with the crest being distinctly U-shaped (3). 82. Shape of temporal fossa: longer anteroposteriorly than wide transversely, or as wide as long (0); wider than long (1). 83. Intertemporal constriction: longer anteroposteriorly than wide transversely (0); as state 1, but with the temporal fossa forming a large parasagittal oval (1); wider transversely than long anteroposteriorly (2). 84. Exposure of frontal on skull vertex: broadly exposed (0); anteroposteriorly compressed or absent (1). 85. Parietal and interparietal: anteriormost point located no further forward than postorbital process (0); anteriormost point in line with supraorbital process (1). 86. Outline of fronto-parietal suture: straight or lobate (0); frontals projects posteriorly along the sagittal plane and separate the left and right parietal anteriorly (1); highly irregular (2). 87. Parietal in lateral view: as long or longer anteroposteriorly than high dorsoventrally (0); higher dorsoventrally than long anteroposteriorly (1).