Oldest evidence for grooming claws in euprimates

Journal of Human Evolution xxx (2018) 1e22

Contents lists available at ScienceDirect

Journal of Human Evolution journal homepage: www.elsevier.com/locate/jhevol

Oldest evidence for grooming claws in euprimates Doug M. Boyer a, *, Stephanie A. Maiolino b, Patricia A. Holroyd c, Paul E. Morse d, e, Jonathan I. Bloch d a

Department of Evolutionary Anthropology, Duke University, P. O. Box 90383, Biological Sciences Building, 130 Science Drive, Durham, NC 27708, USA Department of Anatomical Sciences, Stony Brook University, Stony Brook, NY, USA c Museum of Paleontology, University of California, 1101 Valley Life Sciences Building, Berkeley, CA 94720, USA d Florida Museum of Natural History, University of Florida, Gainesville, FL, USA e Department of Anthropology, University of Florida, Gainesville, FL, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 13 December 2017 Accepted 23 March 2018 Available online xxx

Euprimates are unusual among mammals in having fingers and toes with flat nails. While it seems clear that the ancestral stock from which euprimates evolved had claw-bearing digits, the available fossil record has not yet contributed a detailed understanding of the transition from claws to nails. This study helps clarify the evolutionary history of the second pedal digit with fossils representing the distal phalanx of digit two (dpII), and has broader implications for other digits. Among extant primates, the keratinized structure on the pedal dpII widely varies in form. Extant strepsirrhines and tarsiers have narrow, distally tapering, dorsally inclined nails (termed a ‘grooming claws’ for their use in autogrooming), while extant anthropoids have more typical nails that are wider and lack distal tapering or dorsal inclination. At least two fossil primate species thought to be stem members of the Strepsirrhini appear to have had grooming claws, yet reconstructions of the ancestral euprimate condition based on direct evidence from the fossil record are ambiguous due to inadequate fossil evidence for the earliest haplorhines. Seven recently discovered, isolated distal phalanges from four early Eocene localities in Wyoming (USA) closely resemble those of the pedal dpII in extant prosimians. On the basis of faunal associations, size, and morphology, these specimens are recognized as the grooming phalanges of five genera of haplorhine primates, including one of the oldest known euprimates (~56 Ma), Teilhardina brandti. Both the phylogenetic distribution and antiquity of primate grooming phalanges now strongly suggest that ancestral euprimates had grooming claws, that these structures were modified from a primitive claw rather than a flat nail, and that the evolutionary loss of ‘grooming claws’ represents an apomorphy for crown anthropoids. © 2018 Elsevier Ltd. All rights reserved.

Keywords: Strepsirrhini Eocene Archicebus Omomyiform Exaptation Evolutionary development

1. Introduction Most extant primates have fleshy-tipped fingers and toes with flattened nails (ungulae) instead of sharp claws (falculae). Indeed, the presence of ungulae is often hypothesized to characterize the common ancestor of extant primates, possibly associated with the evolution of manual dexterity (Jones, 1916; Napier, 1961; Le Gros Clark, 1971; Cartmill, 1974; Soligo and Müller, 1999; Soligo and Martin, 2006; Bloch et al., 2007). Exceptions to the ubiquity of these traits include the presence of tegulae (a falcula-like morphology) found in the aye-aye (Daubentonia madagascariensis) and in marmosets and tamarins (Callitrichidae). However, it

* Corresponding author. E-mail address: [email protected] (D.M. Boyer).

is likely that tegulae secondarily evolved from the flattened ungulae more typical among primates (Hamrick, 1998; Soligo and Müller, 1999). Another exception to the ubiquity of ungulae in extant primates is the form of the keratinized structure supported by the distal phalanx of the second digit (dpII) of the foot. In some primates the pedal dpII resembles a dorsally projecting falcula more than a flattened ungula and is referred to as a ‘grooming claw,’ or ‘toilet claw,’ to reflect its observed use in autogrooming and/or scratching (Fig. 1). Groups exhibiting such a ‘grooming claw’ include strepsirrhines (lemurs, lorises, and galagos) and at least the owl monkey (Aotus) and titi monkey (Callicebus) among New World monkeys. In addition, tarsiers have a similar ‘grooming claw’ on both the second and third pedal digits (Soligo and Müller, 1999; Maiolino et al., 2011; Maiolino et al., 2012; von Koenigswald et al., 2012; Maiolino, 2015).

https://doi.org/10.1016/j.jhevol.2018.03.010 0047-2484/© 2018 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

2

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Figure 1. Examples of ‘grooming claw’-bearing distal phalanges and associated ungula-bearing phalanges in strepsirrhine primates. AeB) Foot of Cheirogaleus major (AMNH-M31265) showing ‘grooming claw’ morphology on dpII next to ungula morphology on dpIII in medial (A) and dorsal (B) views. Image derived from MorphoSource media file #M10879-16410. CeE) Grooming phalanges and associated ungula morphology in several individual strepsirrhines: C) Cheirogaleus major (DPC 1285); D) Galago senegalensis (DPC 003); E) Galago senegalensis (DPC 1063F). Scale bar on CeE is 1 mm. Views from top to bottom are lateral, dorsal, ventral, distal and proximal.

Studies such as those by Maiolino et al. (2011) have shown that ungulae, falculae, tegulae and ‘grooming claws’ have underlying distal phalanges with distinctive, correlative morphology. Thus, it is possible to infer which type of keratinized structure was present in extinct primates based on fossilized distal phalanges. Primarily, ungular phalanges are distinguished by dorsoventral flattening, mediolateral expansion, and a distally shifted attachment for the long flexor tendon. Falcular phalanges and tegular phalanges tend to be dorsoventrally expanded, mediolaterally narrow and to have a shaft that is plantarly curved and hook-like (Patel and Maiolino, 2016). In addition, ungular and ‘grooming claw’ phalanges of primates tend to have an ‘apical tuft,’ a fan- or apron-like sheet of bone that rims the distal tip (Maiolino et al., 2011, 2012).

Debates about the appropriate terminology for the ‘grooming claw’ stem from different perspectives on (1) the morphological features that distinguish ‘grooming claw’-bearing distal phalanges (grooming phalanges) from the bony phalanges underlying more typical falculae (falcular phalanges) and ungulae (ungular phalanges), and (2) the homology of these structures (Maiolino et al., 2011; von Koenigswald et al., 2012; Gebo et al., 2015). Understanding the functional, developmental and phylogenetic basis of ‘grooming claw’ morphology contributes to understanding the evolutionary pattern of morphological divergence of euprimates from other euarchontans. More complete understanding of ‘grooming claw’ taxonomic distribution and morphological variation helps constrain hypotheses on the number of times these

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

different morphologies arose and the selective pressures experienced by early primates. Below we briefly review previous perspectives on ‘grooming claw’ taxonomic distribution and homology to provide context and highlight the broader implications of the fossils described here.

3

It has been suggested that ‘grooming claws’ of ‘prosimians’ were inherited from the last common ancestor of modern primates (Soligo and Müller, 1999; Franzen et al., 2009), and that the lack of ‘grooming claws’ in extant anthropoids is likely an apomorphy. The presence of a ‘grooming claw’ in extant Aotus has been used as

Table 1 Fossil specimens discussed here and unique online identifiers for 3D scans on www.MorphoSource.org. Specimens with bold ID are of key interest in this study. Species/size group Teilhardina brandti Teilhardina brandti Teilhardina brandti Teilhardina brandti Teilhardina brandti Teilhardina brandti large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid large omomyid Tetonius matthewi Tetonoides pearcei Arapahovius gazini small omomyid small omomyid small omomyid small omomyid small omomyid small omomyid small omomyid small omomyid small omomyid non-primate non-primate non-primate non-primate non-primate non-primate Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Omomys carteri Notharctus tenebrosus Notharctus tenebrosus Notharctus tenebrosus

Catalogue no.

Locality

Basin

Age

ID

MorphoSource IDa

DOI link

UF 411197 UF 334000 UF 409804 UF 411196 USNM 540587 USNM 521825 UCMP 218398 UCMP 217915 UCMP 218399 UCMP 218416 UCMP 134993 UCMP 217959 UCMP 217922 UCMP 217916 UCMP 218436 UCMP 218368 UCMP 218111 UCMP 217977 UCMP 218247 UCMP 218269 UCMP 218242 UCMP 218157 UCMP 218183 UCMP 218261 UCMP 217999 UCMP 218344 UCMP 218000 UCMP 218432 UCMP 218373 UCMP 218433 UCMP 218310 UCMP 218159 UCMP 218160 UCMP 218244 UCMP 218243 UCMP 218295 UCMP 218379 UCMP 217970 UCMP 218246 UCMP 218245 UCMP 218112 UCMP 218292 DPC 25505 UM 31624-b UM 31624-c UM 31651 UM 32129-a UM 32129-b UM 32129-c UM 32146-a UM 32146-b UM 32146-c UM 31654 UM 32274 UM 32258-a AMNH FM 143612-3 AMNH FM 143612-2 AMNH FM 143612-4

WY06061B WY06061D WY14047b WY06061E WY14047 WY14047 V70215 V70243 V70215 V70220 V70220 V70243 V70243 V70243 V70246 V71231 V74022 V74022 V74022 V74022 V74022 V74022 V74022 V74022 V70214 V71232 V74022 V74022 V70214 V74022 V74022 V74022 V74022 V74022 V74022 V74022 V70214 V70243 V74022 V74022 V74022 V74022 UCM L93026 BRW-14 BRW-14 BRW-14 BRW-14 BRW-14 BRW-14 BRW-14 BRW-14 BRW-14 BRW-15 BRW-35 BRW-139 ALX-00-05 ALX-00-05 ALX-00-05

Bighorn Bighorn Bighorn Bighorn Bighorn Bighorn Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger Bridger

Wa0 Wa0 Wa0 Wa0 Wa0 Wa0 Wa3 Wa3 Wa3 Wa3 Wa3 Wa3 Wa3 Wa3 Wa3 Wa3 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa3 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa5 Wa3 Wa3 Wa5 Wa5 Wa5 Wa5 Br3 Br2 Br2 Br2 Br2 Br2 Br2 Br2 Br2 Br2 Br2 Br2 Br3 Br2 Br2 Br2

Groom Groom Groom Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Groom Groom Groom Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Falcula Falcula Falcula Falcula Falcula Falcula Groom Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Ungula Groom Ungula Ungula

M14822-26655 M10418-15219 M12597-20769 M14815-26647 M10451-15299 M10449-15295 M9257-12526 M9128-12244 M9200-12412 M9260-12535 M9040-12030 M1035-883 M9135-12262 M9129-12247 M15071-27311 M9198-12406 M9151-12302 M9142-12282 M9182-12368 M14750-26506 M9179-12359 M9166-12334 M9176-12354 M17971-34043 M9149-12296 M14094-24761 M9150-12299 M9262-12541 M15072-27312 M9263-12544 M9193-12395 M9168-12338 M9170-12342 M15073-27313 M15074-27314 M17972-34044 M9199-12409 M12978-22115 M9181-12365 M9180-12362 M9152-12305 M15348-28084 M15337-28054 M17973-34046 M17974-34048 M17976-34050 M17981-34052 M17985-34054 M17986-34058 M17987-34060 M17988-34062 M17989-34064 M17990-34066 M17991-34068 M17992-34070 M5695-5136 M5692-5133 M5689-5130

https://doi.org/10.17602/M2/M26655 https://doi.org/10.17602/M2/M15219 https://doi.org/10.17602/M2/M20769 https://doi.org/10.17602/M2/M26647 https://doi.org/10.17602/M2/M15299 https://doi.org/10.17602/M2/M15295 https://doi.org/10.17602/M2/M12526 https://doi.org/10.17602/M2/M28069 https://doi.org/10.17602/M2/M12412 https://doi.org/10.17602/M2/M12535 https://doi.org/10.17602/M2/M12030 https://doi.org/10.17602/M2/M883 https://doi.org/10.17602/M2/M12262 https://doi.org/10.17602/M2/M28072 https://doi.org/10.17602/M2/M12538 https://doi.org/10.17602/M2/M12406 https://doi.org/10.17602/M2/M12302 https://doi.org/10.17602/M2/M12282 https://doi.org/10.17602/M2/M12368 https://doi.org/10.17602/M2/M26506 https://doi.org/10.17602/M2/M12359 https://doi.org/10.17602/M2/M12334 https://doi.org/10.17602/M2/M12354 https://doi.org/10.17602/M2/M38501 https://doi.org/10.17602/M2/M12296 https://doi.org/10.17602/M2/M24761 https://doi.org/10.17602/M2/M12299 https://doi.org/10.17602/M2/M12541 https://doi.org/10.17602/M2/M27312 https://doi.org/10.17602/M2/M12544 https://doi.org/10.17602/M2/M12395 https://doi.org/10.17602/M2/M12338 https://doi.org/10.17602/M2/M12342 https://doi.org/10.17602/M2/M27313 https://doi.org/10.17602/M2/M27314 https://doi.org/10.17602/M2/M38529 https://doi.org/10.17602/M2/M12409 https://doi.org/10.17602/M2/M22115 https://doi.org/10.17602/M2/M12365 https://doi.org/10.17602/M2/M12362 https://doi.org/10.17602/M2/M12305 https://doi.org/10.17602/M2/M28084 https://doi.org/10.17602/M2/M28054 https://doi.org/10.17602/M2/M34046 https://doi.org/10.17602/M2/M34048 https://doi.org/10.17602/M2/M34050 https://doi.org/10.17602/M2/M34052 https://doi.org/10.17602/M2/M34054 https://doi.org/10.17602/M2/M34058 https://doi.org/10.17602/M2/M34060 https://doi.org/10.17602/M2/M34062 https://doi.org/10.17602/M2/M34064 https://doi.org/10.17602/M2/M34066 https://doi.org/10.17602/M2/M34068 https://doi.org/10.17602/M2/M34070 https://doi.org/10.17602/M2/M5133 https://doi.org/10.17602/M2/M5136 https://doi.org/10.17602/M2/M5130

Institutional abbreviations: AMNH ¼ American Museum of Natural History; DPC ¼ Duke Primate Center; UCMP ¼ University of California Museum of Paleontology; UF ¼ University of Florida, Florida Museum of Natural History; UM ¼ University of Michigan, Museum of Paleontology. Locality abbreviations: ALX ¼ John Alexander Locality; BRW ¼ Bridger formation localities assigned by G. F. Gunnell crews; UCM L ¼ University of Colorado Museum locality; V ¼ localities assigned by UCMP; WY ¼ Florida Museum of Natural History localities. a Unique MorphoSource identifier for a scan representing a specimen. b Equivalent to WW-47 “Amy's Hill” that previously yielded assorted elements attributed to Teilhardina brandti by Rose et al. (2011).

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

4

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

evidence against the latter point (Maiolino et al., 2011). Eocene adapiforms (possible stem strepsirrhines) have consistently been found to possess ‘grooming claws’. Fossil species previously known to have grooming claws include Notharctus tenebrosus (Maiolino et al., 2012), Europolemur koenigswaldi (von Koenigswald et al., 2012), Hoanghonius stehlini (Gebo et al., 2015), and Adapoides troglodytes (Gebo et al., 2017). Even as the above cited evidence of ‘grooming claws’ in early fossil species was accumulating, Maiolino, von Koenigswald, Gebo and their collaborators interpreted the data and observations to indicate multiple independent evolutionary origins for the ‘grooming claw’ within euprimates. Maiolino et al. (2012) suggested that the mosaic morphology of the Notharctus ‘grooming claw’ (AMNH FM 143612-3) could indicate that it was in the process of evolving a ‘grooming claw’ from a more primitive condition that lacked ‘grooming claws’. Assuming adapiforms do indeed represent stem strepsirrhines, this would suggest convergence in ‘grooming claw’ morphology between at least strepsirrhines and haplorhines. Similarly, von Koenigswald et al. (2012) asserted that ‘grooming claws’ evolved independently in primates at least three times (once in strepsirrhines, once in tarsiers, and once in the ancestor of Aotus and Callicebus). Ni et al. (2013) reported additional evidence consistent with the view of convergent acquisition of ‘grooming claws.’ They described the foot of the basal haplorhine, Archicebus achilles, and interpreted the pedal dpII as lacking a grooming claw. By implying that ‘grooming claw’ traits evolved multiple times in different lineages from an ungula, these studies question the appropriateness of the term ‘grooming claw’. While the lack of a ‘grooming claw’ in A. achilles would seem to be strong evidence against the homology of ‘grooming claws’ in strepsirrhines and haplorhines, there are problems with the evidence from this specimen. Primarily, the evidence amounts to an impression of the dpII in dorsal view. Maiolino et al. (2011, 2012) have demonstrated that a dorsal view alone is insufficient for diagnosing the presence of a ‘grooming claw’ in extant species known to have one. Therefore, resolution of the form of the pedal dpII in the ancestral euprimate requires additional data from other early primate taxa, particularly from omomyiform primatesdthe radiation of Paleogene primates considered more closely related to extant haplorhines, and particularly tarsiids (Beard et al., 1988; Seiffert et al., 2009; Gingerich, 2012). In this context, if omomyiforms were found to exhibit an ungula-like structure on the pedal dpII (i.e., to lack a grooming claw), this would add convincing support to the hypothesis that ‘grooming claws’ evolved convergently in strepsirrhines and haplorhines. To date, omomyiform distal phalanges have been too sparsely represented to evaluate the form of the pedal dpII. While it is impossible to unequivocally establish the lack of a ‘grooming claw’ in a fossil taxon without articulated material, the diagnostic features of grooming claws do make it possible to demonstrate their presence even from isolated remains. In this study, we describe seven distal phalanges (Table 1) that indicate the presence of ‘grooming claws’ in omomyiform primates. These fossils come from three different time intervals in the early Eocene of

5

Wyoming, USA: (1) Bighorn Basin screen-washing sites in Wa0 (Wasatchian North American Land Mammal Age, Wa0 biozone) exposures of Willwood Formation (Fig. 2A) where Teilhardina brandti dentitions and postcrania are known to occur (Rose et al., 2011); (2) Washakie Basin screen-washing sites in Wa3 to Wa5 exposures of the Wasatch Formation (Fig. 2B), from whence a diversity of omomyiform dentitions have been recovered (Savage and Waters, 1978; Williams and Covert, 1994; Cuozzo, 2002); and (3) the Omomys Quarry site from the Bridger Basin (Fig. 2C), which is Br3 (Bridger North American Land Mammal Age, Br3 biozone) in age (Anemone and Covert, 2000; Murphey et al., 2001, 2017). Three dimensional (3D) scans of the specimens are deposited and available for download and analysis on MorphoSource.org (Boyer et al., 2016), and can be directly accessed through the unique media identifiers and digital object identifier (DOI) links given in Table 1. 1.1. Paleontologic context of new fossils Bighorn Basin grooming phalanges Sorting through screen-wash concentrate from Wa0 localities in the southern Bighorn Basin, P.E.M. and J.I.B. identified three distal phalanges with strong qualitative affinities to the grooming phalanges of extant primates (UF 334000, UF 409804, and UF 411197). The Wa0 biozone has only two species attributed to euprimates: a small omomyiform (T. brandti) and a much larger adapiform (Cantius torresi). The small size of the Wa0 distal phalanges immediately suggested that of these two euprimate candidates, the omomyiform is the most likely. In some regards, these fossil grooming phalanges appear more primitive than the grooming phalanges of extant primates, primarily in retaining a large basal nutrient foramen (Fig. 3). These fossils were collected as part of an ongoing effort led by crews from the University of Florida and Duke University to increase representation of small vertebrate fossils from sites bracketing and sampling the Paleocene-Eocene Thermal Maximum (PETM) in the Bighorn Basin of Wyoming (Wing et al., 2005; Chester et al., 2010; Secord et al., 2012; Baczynski et al., 2013; Bourque et al., 2015). Washakie Basin grooming phalanges Sorting through screen-wash concentrate from Wa3 and Wa5 localities of the Washakie Basin, D.M.B. recovered an additional three specimens (UCMP 217999, 218000, 218344) that appear qualitatively to resemble grooming phalanges of extant primates (Figs. 2 and 3). Generally speaking, primates are extremely abundant from these sites: D.M.B. recovered 27 ungular phalanges (Fig. 2B; Supplementary Online Material [SOM] Table S1) and a total of 395 postcranial specimens attributable to omomyiforms based on their small size and comparisons to previously identified postcranial material of omomyiforms (Szalay, 1976; Savage and Waters, 1978; Dagosto, 1988; Gebo, 1988; Covert and Hamrick, 1993; Dagosto et al., 1999; Hamrick, 1999; Anemone and Covert, 2000; Dunn et al., 2006; Gebo et al., 2012), adapids (Gregory, 1920; Hamrick and Alexander, 1996), and extant primates. An attribution to omomyiforms again seems most likely for the three grooming phalanges. An attribution to adapiforms seems

Figure 2. Fossil phalanges from Willwood, Wasatch, and Bridger Formations, including first grooming phalanges ever attributed to omomyiforms. A) Ungular (left three) and grooming (right three) phalanges from the Bighorn Basin (Wa0) are assigned to Teilhardina brandti on the basis of size and similarity to extant euprimates. From left to right specimens are UF 411196, USNM 521825, USNM 540587, UF 411197, UF 334000, and UF 409804. B) Ungular and grooming phalanges from the Washakie Basin. Dashed line separates bones thought to belong to small species from those belonging to larger species. From left to right specimens are UCMP 218244 (from Wa5), 218243 (from Wa5), 218344 (from Wa3), 218436 (from Wa3), 218157 (from Wa5), 218000 (from Wa5), and 217999 (from Wa3). Grooming phalanges include UCMP 218344 (assigned to Tetonoides pearcei), UCMP 218000 (assigned to Arapahovius gazini, and UCMP 217999 (assigned to Tetonius matthewi). C) Primate distal phalanges from the Bridger Fm. Left two bones represent ungular (UM 32258-a) and grooming phalanges (DPC 25505) attributed to Omomys sp. Right side specimen is the grooming phalanx from an articulated foot of Notharctus tenebrosus (AMNH FM 1436123). See text for further details. Scale bar for all specimens is 2 mm. From top to bottom in each column, views are lateral, dorsal, ventral, distal and proximal (for grooming phalanges, the distal view is to the left of the proximal view).

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

6

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Figure 3. Comparative plate of second distal phalanges in lateral and dorsal view. Phalanges of fossils are labeled with taxonomic attributions determined in this study (for omomyids) and elsewhere (for Notharctus). Phalanges are scaled to the same dorsoplantar depth of the proximal articular surface. Scale bars are 1 mm long. Specimens represented are UM 109683 (Carpoletes simpsoni), UF 411197 (Teilhardina brandti), UCMP 218000 (Arapahovious gazini), UCMP 217999 (Tetonius matthewi), DPC 25505 (Omomys carteri), AMNH FM 143612-3 (Notharctus tenebrosus). Extant taxa are AMNH 37703A (Erinaceus sp.), DPC 1285 (Cheirogaleus major), DPC 003 (Galago senegalensis), DPC 030 (Microcebus murinus), SBU(14) (Lemur catta), AMNH-M-109367 (Tarsius sp.) and SBU(11) (Aotus sp.).

unlikely given the small size of the bones, the low abundance of adapiform fossils at the sites, and apparent morphological differences between the new fossils and previously described adapiform dpII's (Maiolino et al., 2012; von Koenigswald et al., 2012; Gebo et al., 2015). On the other hand, the Washakie Basin grooming phalanges look quite similar to the Bighorn Basin grooming phalanges (including the presence of a large nutrient foramen; Fig. 2). Interestingly, the Washakie Basin phalanx sample includes specimens from differently aged sites: UCMP 217999 (from UCMP locality V70214) and UCMP 218344 (from V7132) are older (Wa3), while UCMP 218000 (V74022) is younger (Wa5). UCMP 217999 and 218000 are essentially the same size, but likely represent different taxa since no large omomyiform species spans the Wa3-5 interval at these localities. UCMP 218344 is substantially smaller, suggesting the presence of a ‘grooming claw’ in a third omomyiform taxon. There are a minimum of two omomyiform taxa at each of the screen-washing sites yielding grooming phalanges (Tables 2 and 3). The Washakie Basin fossils were collected primarily by Don Savage, J. Howard Hutchison, and University of California crews in the 1970s (Savage and Waters, 1978). Fossils were preserved in extensive shell marls and/or fine grained mudstones topping these marls. Stratigraphically- and geographically-constrained quarries were developed in these fossiliferous horizons. Quarries were screen-washed in small batches (shovel scoops) and sorted over

several decades at the University of California, Berkeley. Specimen associations have been maintained to the level of the ‘shovel scoop’ of matrix. Bridger basin grooming phalanges During fieldwork in June 2017, surface collecting around Omomys Quarry (UCM L93026) located in the Hickey Mountain Limestone, Twin Buttes Member of the Bridger Formation (Anemone and Covert, 2000; Murphey et al., 2001, 2017), yielded a bone qualitatively resembling a grooming phalanx of a small primate (DPC 25505; Figs. 2 and 3). Because of this discovery, one of us (S.A.M.) re-examined previouslycollected specimens from Omomys Quarry held at the University of Colorado, but did not recover any additional grooming phalanges. The fieldwork is part of a project starting in 2014 wherein crews led by J.I.B. and D.M.B. have surface prospected and quarried classic sites in Bridger B, C, and D (NSF BCS1440742, 1440588). 2. Materials and methods 2.1. Institutional and collection abbreviations AMNH: American Museum of Natural History, New York, USA; DPC: Duke Lemur Center, Division of Fossil Primates collection, Durham, NC, USA; ECA: the Quercy locality of Escamps, France;

Table 2 Omomyids known from dentitions at fossil sites considered in this study and average lower first molar dimensions. See SOM Table S3 for individual measurements. Taxon Teilhardina brandti Tetonius matthewi Tetonoides pearcei Teilhardina cf. demissa Arapahovius gazini Steinius vespertinus Anemorhysis savagei Omomys carteri

Basin

Locality

Bighorn Washakie Washakie Washakie Washakie Washakie Washakie Bridger

WY06061, WY14047 V70214, V71232 V70214, V71232 V70214, V71232 V74022 V74022 V74022 BRW-14

Biozone

abs Age

Wa0 Wa3 Wa3 Wa3 Wa5 Wa5 Wa5 Br2

~56 ~54 ~54 ~54 ~53 ~53 ~53 ~47

Ma Ma Ma Ma Ma Ma Ma Ma

n

M1 L

M1 W

41 1 9 1 4 1 10 1

1.97 2.20 1.84 1.66 2.43 2.20 1.77 2.54

1.57 2.12 1.44 1.22 2.18 1.80 1.34 2.02

Abbreviations: L ¼ mesiodistal length (in mm); W ¼ buccolingual breadth (in mm).

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

7

Table 3 Dimensions of fossil omomyiform grooming and ungular phalanges. See Table S1 for individual specimen data. Age Wa0 Wa0 ~Wa3 ~Wa3 ~Wa5 Wa3-5 Wa3-5 ~Wa5 ~Wa5 ~Wa3 ~Wa3 ~Br2 ~Br2

Basin

Sample

Site

dpL

SE

dpW

SE

dpH

SE

GM1

GM2

GM3

n

Bighorn Bighorn Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Washakie Green River Green River

omomyid grooms omomyid ungulae UCMP 218344 UCMP 217999 UCMP 218000 large omomyid ungulae small omomyid ungulae large omomyid ungulae small omomyid ungulae large omomyid ungulae small omomyid ungulae DPC 25505 om. ungulae

multiple multiple V71232 V70214 V74022 all all V74022 V74022 multiple V70214 UCM L93026 Multiple

2.25 2.04 e 3.19 3.21 2.31 1.80 2.34 1.79 2.28 1.94 e 2.88

0.10 0.38 na na na 0.06 0.04 0.08 0.04 0.09 na na 0.08

0.83 1.02 0.93 1.10 1.20 1.27 1.05 1.29 1.08 1.26 0.93 1.92 1.89

0.01 0.06 na na na 0.05 0.03 0.06 0.03 0.08 na na 0.03

0.86 0.55 0.92 1.18 1.18 na na na na na na 1.86 1.01

0.02 0.08 na na na na na na na na na na 0.03

1.17 1.04 e 1.61 1.66 na na na na na na e 1.77

0.85 0.75 0.92 1.14 1.19 na na na na na na 1.89 1.38

1.37 1.44 na na na 1.71 1.36 1.74 1.39 1.70 1.34 na 2.34

3,3,3 3,3,3 0,1,1 1,1,1 1,1,1 18,14,0 9,9,0 9,6,0 7,7,0 9,8,0 1,1,0 0,1,1 11,11,11

Abbreviations: dp ¼ distal phalanx; GM ¼ geometric mean (see text for differences between GM#1, #2 and #3); H ¼ dorsoventral height of proximal end (in mm); L ¼ maximum proximodistal length (in mm); W ¼ mediolateral width of proximal end (in mm); n ¼ sample sizes for each measurement (given respectively); na ¼ not available; SE ¼ standard error (standard deviation divided by square root of sample size). See Table 1 footnote for other abbreviations.

FLMNH: Florida Museum of Natural History, Gainesville, FL, USA; FM: Fossil Mammals division of AMNH collection; M: Mammalogy division of AMNH collection; SBU: Stony Brook University, Stony Brook, NY, USA; UCM: University of Colorado Museum of Natural History, Boulder, CO, USA; UCMP: University of California Museum of Paleontology, Berkeley, CA, USA; UNSM: University of Nebraska Science Museum, Lincoln, NB, USA; USNM: Smithsonian Institution National Museum of Natural History (formerly U.S. National Museum), Washington, DC, USA; YPM: Yale Peabody Museum, New Haven, CT, USA.

analysis, and the sizes of the fossil distal phalanges were subsequently compared to the size of fossil omomyiform teeth. Our goal was to assess if the fossil combinations fit the euprimate or the erinaceid relationship. The sizes (length and width) of the fossil grooming phalanges were also compared to those of ungular phalanges from the same sites in terms. Finally, the new data were entered into a phylogenetic analysis and used in character mapping to evaluate the evolutionary history of the pedal dpII among primates. 2.3. Morphometric comparative analysis

2.2. Diagnostic features of ‘grooming claws’ Recent studies have established diagnostic morphologies characterizing bony phalanges that underlie ‘grooming claws’ in extant taxa (Maiolino et al., 2011, 2012; von Koenigswald et al., 2012; Maiolino, 2015), making it possible to tentatively identify isolated distal phalanges as ‘grooming claw’-bearing (Gebo et al., 2015). However, there are a number of non-primate species that also have ‘grooming claws’ (Maiolino, 2015) and multiple species of primates are known to have coexisted at some of the localities yielding grooming phalanges. Therefore, quantitative data capturing complex morphological variation, as well as absolute size variation is useful for rigorously determining taxonomic attributions of isolated fossils. A workflow was designed to (1) evaluate initial qualitative impressions of fossil distal phalanges as grooming phalanges, and to (2) attribute them to particular species. Comparative morphometric analyses formed the basis for testing the identification of fossil distal phalanges as grooming phalanges. Two different sets of multivariate analyses were performed with independently derived metrics. In all cases, these metrics were derived from digitized versions of the bones: either via 2D photographs or 3D microCT scans. The comparative sample for each analysis was unique, but both shared specimens in common. Both samples consist of euprimate ungular phalanges, euprimate grooming phalanges, non-primate falcular phalanges, plesiadapiform falcular phalanges, extant erinaceid falcular and grooming phalanges, and five or six fossils qualitatively identified as euprimate grooming phalanges from the Willwood, Wasatch, and Bridger Formations. Following confirmation that fossil distal phalanges have their strongest shape similarities to grooming phalanges of extant primates through multivariate shape analyses, the species-level taxonomic attributions of these bones were assessed. The relationship of dental to distal phalanx measurements for extant primates and erinaceids was characterized through regression

Multivariate analyses of measured variables The comparative sample comprised 250 pedal distal phalanges from extant (n ¼ 226) and fossil (n ¼ 24) taxa. The extant sample includes 38 non-primate falcular phalanges, 29 non-primate grooming phalanges, 77 primate ungular phalanges, and 82 primate grooming phalanges (SOM Table S2). Falcular and ungular phalanges were sampled primarily from the third and fourth rays. Grooming phalanges (both non-primate and primate) were from the second and third (when applicable) rays. The fossil sample includes 11 ungular phalanges attributed to omomyiforms from the Bighorn, Washakie, or Bridger Basins; 6 potential fossil euprimate grooming phalangesdthe focal specimens of the analysis; 3 nonprimate falcular phalanges from the Wasatch Formation of the Washakie Basin; and 3 phalanges of N. tenebrosus of known digit position from the associated foot of AMNH 143612 (Maiolino et al., 2012; SOM Table S3). A set of 15 measurements was collected from each specimen. Measurements (Fig. 4; Table 4; SOM Table S4) were either taken from photographs or screenshots of 3D reconstructions of microCT scans using the ruler tool in Adobe® Photoshop® CS6 Extended Edition. Linear measurements were taken in millimeters, angular in degrees. All non-angular measurements were converted to sizeadjusted shape variables through division by their geometric mean (¼ the nth root of the product of n measurements). All angular measurements were converted to radians. Each sizeadjusted shape variable and angular measurement was converted into a z-score by subtracting the variable's mean and dividing by the variable's standard deviation. This standardization created a set of variables with a mean of 0 and a standard deviation of 1 within each variable for use in ordination methods that expect variables to have similar variances. Finally, separate means were calculated for each distinct distal phalanx type within each species (i.e., separate means were calculated for grooming phalanges and ungular phalanges for species that have both). Nearest neighbors to the fossil

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

8

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Figure 4. Illustration of measurements used in manual analyses on a grooming phalanx of Lemur catta, SBU (14). See Table 4 and SOM Table S4 for details and abbreviations. Calculations of angular measurements SIA, FIA, and FSA are shown at bottom of figure.

Table 4 Abbreviations of measurement variables. See SOM Table S4 for detailed descriptions. Abbreviations FH ETH FTH MPL VPL MSH ATH BW MSW ATW WSM WIM SIA FIA FSA

Full name Facet height Extensor tubercle height Flexor tubercle height Maximum phalangeal length Volar process length Midshaft height Apical tuft height Base width Midshaft width Apical tuft width Width of superior margin of articular facet Width of inferior margin of articular facet Included angle of shaft Included angle of articular facet Facet-shaft angle

specimens were calculated as the Euclidean distances using zscores of all variables. Among the potential fossil grooming phalanges, two were slightly damaged (UCMP 217999 and 218000) and one was badly damaged (DPC 25505; Fig. 2). The more subtly damaged bones have slightly abraded bases and are missing small distal portions of their shafts. To assess if this damage substantially affected results, measurements of incomplete regions were taken from versions of the specimens where the missing regions had been reconstructed.

The measurements based on these reconstructions were then analyzed in tandem with measurements taken on the original specimens (MPL, BW, and WSM for UCMP 217999; and MPL, BW, and WIM for UCMP 218000, see Fig. 4, Table 4 for measurement abbreviations, and SOM Table S4 for measurement descriptions). We used the following protocol in reconstructing missing tips of UCMP 217999 and 218000: The distal-most point of the tip was reconstructed as the point of intersection between a straight line following the dorsal contour of the shaft and a straight line following the volar contour of the shaft in lateral view. The missing portion of the base was reconstructed by mirror imaging the undamaged side (SOM Fig. S1). We deemed the breakage of DPC 25505 too extensive to either (a) measure broken surfaces as if they were complete, or (b) attempt to reconstruct morphology with the above protocol. Instead, we analyzed this bone using a reduced measurement dataset as described below. Data were analyzed in R v3.0.3 (R Core Team, 2014). Principal components analyses (PCA) were run using the ‘princomp’ function in the base stats package of R. Pearson's correlation coefficients between each variate and each variable were used to determine loadings (using the ‘corr’ function in the base stats package). Fossil specimens and species mean values for extant distal phalanges were compared to determine whether the fossils exhibit morphology consistent with primate grooming phalanges. Two PCAs were run, one using all 15 measurements, and one using the subset of 11 measurements (FIA, FSA, FH, ETH, FTH, VPL, MSH, BW, MSW, WSM, WIM; see Fig. 4, Table 4, and SOM Table S4) that can be

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

taken from DPC 25505. Note that exclusion of some measurements necessarily altered the geometric means used in converting linear measurements to z-scores for the PCA using the subset of measurements compared to those used for the PCA of all measurements (geometric mean of all linear measurements). Automated geometric analysis In contrast to linear and angular measurements traditionally used to describe shape, computer graphics researchers and applied mathematicians have recently developed techniques for automatic whole surface quantitative comparison of virtual objects, usually representing manufactured structures (Kim et al., 2012), but sometimes organic ones (Styner et al., 2006; Cates et al., 2008; Boyer et al., 2011, 2012; Paniagua et al., 2012). Objects are ‘virtualized’ through 3D scanning methods such as CT, laser scanning or photogrammetry, or they are ‘built’ in software like CAD and Blender. We implemented the ‘auto3dgm’ method (Boyer et al., 2015) in this study, which has already been used in a number of comparative morphology studies (Seiffert et al., 2015; Boyer et al., 2017; Vitek et al., 2017). Like other automated methods, ‘auto3dgm’ requires 3D models of bones for its analyses. To generate 3D models, distal phalanges were microCT-scanned at Duke University's Shared Materials Instrumentation Facility, Stony Brook University's Center for Biotechnology and University of Florida's Nanoscale Research Facility. The CT volumes were segmented in Avizo 8.1 or higher (FEI Visualization Sciences Group, 2009) to generate surface models. Surface models were further processed in Geomagic (3D Systems Inc., 2013) to remove noise, such as stray points and polygons, or points and polygons representing internal features. Scans were made at a resolution yielding sufficiently detailed surface mesh models of between 150,000 and 1,000,000 faces. Raw scan data and derived surface models from fossils used in this analysis are available through MorphoSource.org (Table 1). All further steps in the analysis were performed within the program ‘auto3dgm’ (see workflow below). The benefit of this type of approach compared to manually taking measurements is the increased fidelity and completeness of surface quantification, and the removal of observer error and bias in landmark selection and identification (Boyer et al., 2015). With a more complete representation of surface geometry, this method presents the potential for highlighting important aspects of variation not captured by analyses of researcher-collected measurements. Our protocol for the ‘auto3dgm’ analysis is as follows: smoothed/cleaned surface files were downsampled to a subset of 1100 evenly spread coordinates (called ‘pseudolandmarks’). Rough alignments based on the major axes of these point clouds were then refined through an iterative process whereby local pseudolandmark correspondences are reassigned and the surfaces are rotated in an effort to minimize the Procrustes distance between them. This procedure was repeated for every pair of surfaces in the sample and the resulting Procrustes distance matrix was used to define a minimum spanning tree (MST). The path of the MST was used to propagate pairwise correspondences throughout the dataset and thereby automatically construct a global pseudolandmark dataset of 1100 points representing each surface. Prior to the MST step, pairwise comparisons of very different objects can be expected to produce wrong alignments. For example, if attempting to directly align a falcular phalanx with an ungular phalanx, the dorsoventral axis of the falcular phalanx can be reasonably expected to align with the mediolateral axis of the ungular phalanx. However, imposing the MST while assigning correspondences among different surfaces ensures that only direct alignments of very similar shapes are used. Thus, all objects usually become correctly orientated with respect to each other after this step. Finally, this global correspondence dataset was output as a Morphologika2.5

9

(O'Higgins and Jones, 1998) file. The Morphologika2.5 file was analyzed and visualized using PCA in Morphologika2.5 and PAST (Hammer et al., 2001). The sample for the automated analysis (SOM Table S5) was slightly different from that used for the analyses of measurement variables (SOM Tables S2eS3), because 3D scans were not available for all specimens used in the manual analysis. The presence/absence of the nutrient foramina of the distal phalanges also posed a challenge for the automated shape analysis due to the often variable and taxonomically undiagnostic distribution of these geometrically prominent features. Furthermore, we had a hard time capturing their presence reliably on 3D rendered surfaces. Difficulty in capturing foramen geometry occurs when they are filled in with soft tissue or rock matrix. Although ungular phalanges tend to lack these foramina, whereas falcular phalanges do not, this dichotomy is not universal. Therefore, prior to analysis, the area around the nutrient foramina was digitally covered and smoothed. 2.4. Taxonomic attribution After determining the morphological affinities of the fossils, we used two different methods to make taxonomic assignments of the focal specimens. First, least squares regression was used to regress tooth size against distal phalanx size, in the extant sample. Then we checked whether the fossil distal phalanges predict the size of any omomyiform teeth recovered from the same localities and whether there are any other non-primate (erinaceomorph) species with teeth of the same size. Second, the ratio of third or fourth pedal ungular phalanx size to the grooming phalanx size in living primates was computed, and these values were compared to the size ratio of fossil ungular to fossil grooming phalanges from the same localities. The comparative samples for establishing the relationships between grooming phalanx dimensions, tooth size, and ungular dimensions consisted of 13 genera and 18 species representing strepsirrhines, tarsiers and Aotus, as well as a sample of erinaceomorphs including two extant and one extinct species (SOM Tables S6eS8). The fossil erinaceid is a European species of Macrocranion from Messel, for which measurements on associated distal phalanx lengths and dentition were available for two specimens from the literature (Maier, 1979). These regressions and ratios utilize three measurements of distal phalanges. Ungular phalanges were quantified by distal phalanx total length and distal phalanx proximal end width, which were combined as Geometric Mean #3 (GM3) in Table 3 (individual specimen values appear in SOM Table S1). Grooming phalanges were quantified either by the geometric mean of length, width and proximal end depth (Geometric Mean #1, GM1), or geometric mean of just width and depth (Geometric Mean #2, GM2). These geometric means were used to construct ratios of grooming phalanx size to ungular phalanx size. Two different regressions were run to establish the ability of grooming phalanx size to predict tooth size (see SOM Tables S6eS8 for regression data). In one analysis, the natural logarithm (ln) of molar area (length  width) was regressed against ln distal phalanx articular surface area (width  depth). In a second analysis, ln molar length was regressed against the ln distal phalanx length. 2.5. Cladistic analysis and character optimization A recently published character matrix with comprehensive representation of extant and fossil primates and their skeletal morphology (Ni et al., 2016) was used to assess the phylogenetic context of the new fossil and comparative distal phalanx morphology. Two characters in this matrix pertain to the ‘grooming

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

10

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

claw’ of pedal digit two, with one of them requiring modification. The original coding for character #1009 (#1008 for tnt file starting with character #0) ‘presence of toilet claw’ was ‘absent’ (0), ‘present on one digit’ (1), or ‘present on two digits’ (2). This is problematic, since an isolated grooming phalanx does not reveal the number of grooming phalanges per foot. Furthermore, this coding scheme does not distinguish between falculae and ungulae occurring on pedal digit II. Another character (#1007 ¼ #1006 for tnt file starting with character #0) codes for the presence/absence of ‘deep and mediolaterally compressed ungual phalanges,’ but it is not clear if this character and character #1009 together would always provide sufficient information to distinguish falcula from ungula on the second pedal digit, specifically. Without separate states for falculae and ungulae in character #1009, the morphological evolution of the pedal dpII cannot be fully mapped and evaluated. Therefore we modified character #1009 to reflect the presence of an ungular phalanx (0), grooming phalanx (1), or falcular phalanx (2). Taxon codings in the matrix were modified accordingly. Character #1008 codes for the form of pedal digit III, with one of the character states (4) specifying presence of a ‘toilet claw,’ and therefore the unique condition of tarsiers (with a ‘grooming claw’ on dpII and dpIII) is still captured within the analysis. The codings for all operational taxonomic units (OTUs) under the modified version of character #1009 are indicated in SOM Figure S2. For several OTUs, character states used by Ni et al. (2016) were changed to reflect different interpretations that we believe to be supported by stronger evidence. In the original matrix, Ni et al. (2016) coded N. tenebrosus and D. madagascariensis as ‘0’ for character #1009, indicating lack of a 'grooming claw'. However the presence of a ‘grooming claw’ in both of these taxa has now been well established (Soligo and Müller, 1999; Maiolino et al., 2012; von Koenigswald et al., 2012; Maiolino, 2015). Furthermore, the presence of a ‘grooming claw’ in H. stehlini was coded as ‘?’ However, Gebo et al. (2015) made a strong case for the presence of a ‘grooming claw’ in H. stehlini. The coding for modified character #1009 in these three taxa was accordingly changed to ‘1’ (SOM Fig. S2). The coding of one of the outgroup taxa, Erinaceus sp., also required modification. It was originally coded as state ‘0’ or ‘lacking a grooming claw,’ but recent comparative observations (Maiolino, 2015) and the morphometric findings of this study (see Results section) strongly indicate the presence of distinctive ‘grooming claw’ morphology on dpII in these taxa. Finally, while the early haplorhine A. achilles was coded as lacking a ‘grooming claw’ by the original authors, positive evidence for this designation is lacking and its coding was here changed to a question mark: In the single specimen of Archicebus, only a dorsal view of the impression left from the obliterated digit II pedal phalanx is available (Ni et al., 2013). As mentioned above, a dorsal view is insufficient for distinguishing a ‘grooming claw’ from an ungula among primates (Maiolino et al., 2011, 2012; von Koenigswald et al., 2012). Character #1010 (#1009 for tnt file starting with character #0), ‘shape of the grooming claw,’ in the Ni et al. (2016) matrix was also coded for the new fossil specimens. This character has two states: ‘elongated shaft with expanded rim’ (0), and ‘triangular, wedgeshaped’ (1). In the Ni et al. (2016) matrix, only tarsiers were coded with state ‘1.’ Codings for the recently described talus of Donrussellia provincialis were also included, using character states published in that study (Boyer et al., 2017). All codings for the fossil specimens can be found in the nexus file supplied as SOM S1. A cladistic analysis of the modified character matrix was performed in TNT (Goloboff et al., 2003, 2008; Goloboff and Catalano, 2016) using the settings and scripts provided by Ni et al. (2013, 2016); we also provide the commands in this script as SOM S2. After recovering the most parsimonious trees and computing a strict consensus, character #1009 was optimized over the tree to

assess whether the presence of a ‘grooming claw’ on pedal digit two represents the primitive state for primates or not. 3. Results The results are organized into several sections below. In the first section, we describe results of several different ways of comparing shapes among samples of distal phalanges. The second section evaluates which fossil species known from teeth at the sites yielding the grooming phalanges may be the owners of the grooming phalanges. The third section evaluates the phylogenetic significance of the results. 3.1. Morphometric comparisons Multivariate morphometrics of measured variables Two principal components analyses (PCA) using different sets of manually collected measurements were executed. The first included 15 linear and angular measurements (Fig. 4; Table 4 and SOM Table S7), while the second included a subset of those measurements that could be taken on the most damaged bone DPC 25505 (FIA, FSA, FH, ETH, FTH, VPL, MSH, BW, MSW, WSM, WIM). The first two principal components (PC) of the PCA that included all measurements account for 62% of the total variance (45% and 17% respectively; Fig. 5A). PC1 clearly distinguishes ungular phalanges of extant primates from distal phalanges of extant nonprimates. However, non-primate grooming phalanges are not well distinguished from falcular phalanges. The variables most strongly correlated with PC1 are widths (BW and ATW), indicating that primate ungular phalanges tend to be relatively wider than those of non-primates in this study. Primate grooming phalanges are intermediate in this regard. PC2 distinguishes primate grooming phalanges from all other groups (Fig. 5A). The variables most strongly correlated with this component are the two angles, FSA and FIA, and MPL. Primate grooming phalanges tend to have relatively low values for FSA, meaning their shafts are more dorsally canted, low values for FIA, meaning their articular facets are less concave (and in some cases even convex), and high values for MPL, meaning they are relatively long. The three fossil non-primate specimens from the Washakie Basin fall with extant nonprimates, while all other fossils fall with or near extant primate groups. The focal specimens of this study plus the grooming phalanx of Notharctus all fall within or very near to the space occupied by extant primate grooming phalanges with one exception: UF 334000 from Wa0 falls between grooming phalanges and nonprimate phalanges (‘Tbc’ in Fig. 5). The first two components of the PCA containing the subset of measurements accounted for 64% of the total variance (44% and 20% respectively; Fig. 5B). Results are similar to the first analysis; however, there is some overlap of non-primate phalanges and extant primate grooming phalanges, most likely due to the absence of MPL in this analysis. The focal specimens (including DPC 25505) all fall within the space occupied only by extant primate grooming phalanges. Lastly, ‘nearest neighbors’ were calculated as Euclidean distances among specimens using z-scores of all linear and angular variables for the focal fossils (except for DPC 25505, for which the distances were calculated based on the aforementioned subset of 11 variables). The nearest (non-fossil) neighbor of each of the focal specimens are grooming phalanges of extant primates, with the exception of the fragmentary specimen DPC 25505 (Table 5). Automated morphometrics PCA of the 1100 pseudolandmark coordinates (SOM Table S7) on 67 distal phalanges showed distinct partitioning of phalanges by their functional type (i.e., claw, nail, grooming claw; Fig. 6). The percentage variance represented by

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

11

Figure 5. Plot of the first two principal component scores (PC2 vs. PC1) for distal phalanges derived from principal component analyses (PCA) of linear and angular measurements for visualization of shape differences among groups representing different digit shapes. Each point in the plot represents an extant species mean for a given ungual type or an individual fossil specimen. Open symbols represent dpII that sported grooming claws for extant taxa (as well as tarsier dp3), or that appeared qualitatively to resemble grooming phalanges for fossil taxa. Black-filled symbols represent falcula for non-primates and ungulae for primates. Grey-filled symbols represent grooming claws with reconstruted features. A) PCA including all measurements, but excluding DPC 25505. B) PCA including only measurements that could be taken on incomplete specimen DPC 25505. See Materials and methods, SOM Figure S1, and SOM Tables S5eS9 for more details. Abbreviations: Ag ¼ Arapahovius gazini (UCMP 218000); M ¼ claws from indeterminate non-primate mammals of the Washakie Basin (Ma ¼ UCMP 218379; Mb ¼ UCMP 218112; Mc ¼ UCMP 218292); Nt ¼ Notharctus tenebrosus (nail: Ntn ¼ AMNH-FM-143612-2 and 4; grooming claw: Ntg ¼ AMNH-FM-143612-3); Oc ¼ Omomys carteri (nail: Ocn ¼ UM 32258_a; grooming claw: Ocg ¼ DPC 25505); Tb ¼ Teilhardina brandti (nail: Tbn ¼ USNM 540587; grooming claws: Tbc ¼ UF 334000; Tbd ¼ 409804; Tbe ¼ 411197); Tm ¼ Tetonius matthewi (grooming claw: UCMP 217999).

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

12

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Table 5 Nearest neighbor lists for fossil distal phalanges of interest, derived from manual analyses. Calculated as Euclidean distances for all variables. Bold cells are non-primates. Specimen/taxon UF 409804 Teilhardina brandti UF 334000 Teilhardina brandti UF 411197 Teilhardina brandti UCMP 217999 Tetonius matthewi UCMP 217999a Tetonius matthewi UCMP 218000 Arapahovius gazini UCMP 218000a Arapahovius gazini DPC 25505b Omomys carteri

nn1

nn2

nn3

nn4

nn5

nn6

UF 334000 Teilhardina brandti UF 411197 Teilhardina brandti UF 334000 Teilhardina brandti UCMP 217999a Tetonius matthewi UCMP 217999 Tetonius matthewi UCMP 218000a Arapahovius gazini UCMP 218000 Arapahovius gazini P2 mean (gc) Hemiechinus auritus

UF 411197 Teilhardina brandti UF 409804 Teilhardina brandti UF 409804 Teilhardina brandti P2 mean (gc) Galago senegalensis P2 mean (gc) Galago senegalensis UCMP 217999 Tetonius matthewi UCMP 217999 Tetonius matthewi P2 mean (gc) Cheirogaleus medius

UCMP 218000a Arapahovius gazini UCMP 217999a Tetonius matthewi UCMP 218000a Arapahovius gazini UCMP 218000 Arapahovius gazini P3 mean (gc) Tarsius pelengensis P2 mean (gc) Galago senegalensis P2 mean Galago senegalensis P2 mean (gc) Lepilemur leucopus

P2 mean (gc) Perodicticus potto UCMP 217999 Tetonius matthewi UCMP 217999a Tetonius matthewi P3 mean (gc) Tarsius pelengensis P2 mean (gc) Galago moholi P3 mean (gc) Tarsius bancanus P3 mean Tarsius syrichta P2 mean (gc) Lemur catta

P2 mean (gc) Galagoides demidovii UCMP 218000a Arapahovius gazini UCMP 217999 Tetonius matthewi UCMP 218000a Arapahovius gazini. P2 mean (gc) Tarsius pelengensis P2 mean (gc) Galago moholi UCMP 217999* Tetonius matthewi P2 mean (gc) Arctocebus calabarensis

P3 mean (gc) Tarsius pumilus P3 mean (gc) Tarsius pelengensis P2 mean (gc) Perodicticus potto P2 mean (gc) Galago moholi UCMP 218000 Arapahovius gazini P3 mean (gc) Tarsius syrichta P3 mean Tarsius bancanus P4 mean (f) Hemiechinus auritus

Abbreviations: gc ¼ grooming claw; f ¼ falcula; nn ¼ nearest neighbor. a Reconstructed fossil. b DPC 25505 nearest neighbors are calculated based only on the abbreviated list of variables that could be measured on the damaged specimen.

Figure 6. Plot of the first three principal component (PC) scores generated in Morphologika2.5 for distal phalanges derived from principal component analysis (PCA) of 1100 automatically generated correspondence points per bone (following Boyer et al., 2015). X-axis is PC1 in both A and B. Each point in these plots represents an individual specimen for both extant and fossil taxa. Axes are scaled to the proportion of variance they represent. Open symbols represent dpII that sported ‘grooming claws’ for extant taxa (as well as tarsier dp3), or that appeared qualitatively to resemble grooming phalanges for fossil taxa. Filled symbols represent falcula for non-euprimates and ungulae for euprimates. Among noneuprimates, open and black-filled squares represent erinaceids, dark gray squares represent fossil plesiadapiforms and light gray squares represent treeshrews. Abbreviations: Ag ¼ Arapahovius gazini (UCMP 218000); Ap ¼ Adapis parisiensis (ECA-1400); Nt ¼ Notharctus tenebrosus (grooming claw: AMNH-FM-143612-3); om, omomyid ungulae from Washakie localities (oma ¼ UCMP 218244; omb ¼ 218436); Tb ¼ Teilhardina brandti (grooming claws: Tbc ¼ UF 334000; Tbd ¼ 409804; Tbe ¼ 411197; ungulae: Tbb ¼ UF 411196; Tba ¼ USNM 521825); Tm ¼ Tetonius matthewi (UCMP 217999). Asterisk on ‘Tm’ indicates a version of the specimen (UCMP 217999) for which the tip was digitally reconstructed. We present this version and the broken version as a suggestion of how much an effect the breakage has on the shape analysis. UCMP 218000 is broken in a similar way. Circles with an asterisk to the left are grooming phalanges of Aotus.

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

the first three principal components amounted to 60% (PC1 ¼ 33%; PC2 ¼ 19%; PC3 ¼ 8%), while the variance represented by each additional axis plummeted on PC4 and higher. Compared to the manually collected measures, the automated analysis results suggest a more nuanced view of distal phalanx morphology. PC1 clusters bones along a spectrum from flattened ungular phalanges (lowest PC1 scores; Fig. 6) to strepsirrhine and anthropoid grooming phalanges, to erinaceid grooming phalanges, to tarsier grooming phalanges, to falcular phalanges and erinaceid lateral distal phalanges (highest PC1 scores; Fig. 6). On PC2, primate ungular phalanges and erinaceid lateral distal phalanges have higher values than everything else, probably reflecting low overall shaft length relative to width (Fig. 6A). On PC3, tarsier grooming phalanges are distinguished from everything else (Fig. 6B). The fossil dpII of Notharctus (AMNH FM 143612-3) plots with other strepsirrhine grooming phalanges, as in previous studies. A scutiform distal phalanx attributed to Adapis (ECA 1400; Godinot, 1992) plots far away from other ungular phalanges. Nonetheless, its position in PC morphospace reflects the previous observations (Godinot, 1992) that it retains some claw-like morphologies (high PC1 score) as well as an extremely proximodistally short shaft relative to its dorsoventral depth and mediolateral width (high PC2 score). Finally, the newly recovered grooming phalanges plot in a variety of positions with respect to extant taxa. UCMP 217999 seems unambiguously closest to extant strepsirrhine grooming phalanges. Its nearest neighbor with respect to the Procrustes distance of 1100 pseudolandmarks is a specimen of Galago senegalensis (Table 6). UCMP 218000, on the other hand, seems perhaps closer to erinaceid grooming phalanges, though its nearest neighbors are the other fossil omomyiform grooming phalanges identified in this study (Table 6). Finally, the Wa0 specimens likely representing T. brandti grooming phalanges all plot closest to tarsier grooming phalanges and have tarsier grooming phalanges for their first nearest extant neighbors (Table 6). Univariate and bivariate assessment Although a multivariate approach allows for maximal separation among a sample with many different distal phalanx types, we found that only two variables are needed to distinguish primate grooming phalanges from other phalanx types (Fig. 7; Table 7). These two variables include the facet-shaft angle (FSA) and the included angle of the articular facet (FIA). While similar FSA values show that both non-primate and primate grooming phalanges have a dorsally projecting shaft (probably the most ‘functionally’ relevant distinguishing feature), their divergent FIA values show that non-primate grooming phalanges tend to have much more concave articular facets (Fig. 3; Erinaceus). Thus, taking note of these two features, there is probably little chance of misidentifying a non-primate for a primate grooming phalanx, despite the overlap in some of our multivariate analyses. All fossil primate grooming phalanges

13

share this bivariate shape space with extant primate grooming phalanges (Fig. 7C). Finally, though multivariate morphometrics of measured variables did not identify consistent differences between tarsier and strepsirrhine grooming claws, automated morphometrics did. Thus, we wondered if there were any specific measured variables separating these two groups. Qualitatively, all of the omomyid grooming phalanges appeared to have more pronounced flexor tubercles; tarsiers seemed qualitatively similar in this feature. Computing the ratio of facet area √(FH*BW) to the flexor tubercle height (FTH) reveals indeed that certain tarsiers tend to have a more pronounced plantar projection of the flexor tubercle (Table 6); however, they are not significantly distinguished from grooming claws of any major strepsirrhine groups when taking into account intraspecific variation. On the other hand, the newly described omomyid grooming phalanges are quite distinct from all of the extant grooming claws (except several aforementioned tarsier specimens) in their flexor tubercle projection, as well as from the grooming phalanx of Notharctus (Table 7). In this regard, the omomyid grooming phalanges approach the condition of falcular and erinaceid phalanges more closely than any of the extant primate bones. 3.2. Taxonomic attribution To further test the attribution of the new fossil grooming phalanges to euprimates generally, we evaluated whether these fossils are the correct size for attribution to any of the specific primate dentitions or ungular distal phalanges recovered from the same localities. Sizes of fossil grooming phalanges were compared to those of previously identified ungular phalanges within each locality. Extant primates were examined to understand how close in size grooming and ungular phalanges from a single primate species typically are. For extant taxa (a sample of 15 strepsirrhines, three tarsiers, and two owl monkeys), ratios of ungular to grooming phalanx size ranged from 0.82 to 1.01 (Table 8), meaning that an individual's grooming and ungular phalanges tend to be similar in size. We note that an individual's grooming phalanges are almost always longer and therefore visually appear bigger than its ungular phalanges. However, their overall ratios are similar because grooming phalanges are narrower than ungular phalanges. Ungular phalanx lengths in the Wasatch Fm. sample occur in two size classes: those slightly larger than 2 mm (n ¼ 19) and those slightly shorter than 2 mm (n ¼ 11; Tables 1 and 2, and SOM Table S1). Mann-Whitney-U test on these samples showed that their lengths are significantly different at p < 0.01. When the geometric means of the grooming phalanges were compared to the ungular phalanges from the same locality, the ratio of the average geometric mean of UCMP 217999 and 218000 to the

Table 6 Nearest neighbor lists for automated geometric analysis of 1100 pseudolandmark coordinates. Bold cells are non-primates. Specimen/taxon UF 409804 Teilhardina brandti UF 334000 Teilhardina brandti UF 411197 Teilhardina brandti UCMP 217999 Tetonius matthewi UCMP 218000 Arapahovius gazini

nn1

nn2

nn3

nn4

nn5

nn6

UF 334000 Teilhardina brandti UF 409804 Teilhardina brandti UF 409804 Teilhardina brandti DPC 003 Galago senegalensis UCMP 217999 Tetonius matthewi

UF 411197 Teilhardina brandti UF 411197 Teilhardina brandti UCMP 218000 Arapahovius gazini SBU-PGa1163 Otolemur crassicaudatus UF 411197 Teilhardina brandti

AMNH M 106754 Tarsius bancanus AMNH M 106754 Tarsius bancanus AMNH M 106754 Tarsius bancanus UCMP 218000 Arapahovius gazini AMNH 185374A Hemiechinus sp.

AMNH 109367 Tarsius spectrum UCMP 218000 Arapahovius gazini UF 334000 Teilhardina brandti SBU-15 Galago senegalensis SBU-PGa1163 Otolemur crassicaudatus

UCMP 218000 Arapahovisu gazini USNM 196477 Tarsius pumilus UCMP 217999 Tetonius matthewi DPC 024 Otolemur crassicaudatus DPC 024 Otolemur crassicaudatus

USNM 196477 Tarsius pumilus AMNH 109367 Tarsius spectrum AMNH 109367 Tarsius spectrum DPC 097 Mirza coquereli AMNH 185374B Hemiechinus sp.

Abbreviations: gc ¼ grooming claw; nn ¼ nearest neighbor.

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

14

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Figure 7. Measurements that most strongly distinguish primate grooming claws from other distal phalanges. A) Boxplots of z-scores of FIA compared to individual values for focal fossils. B) Boxplots of z-scores of FSA compared to individual values for focal fossils. C) Bivariate plot of FSA against FIA. Coloration of plots follows that of Figure 3. For raw values see Table 4. Abbreviations: Ag ¼ Arapahovius gazini (UCMP 218000); M; claws from indeterminate non-primate mammals of the Washakie Basin (Ma ¼ UCMP 218379; Mb ¼ UCMP 218112; Mc ¼ UCMP 218292); Nt, Notharctus tenebrosus (nail: Ntn ¼ AMNH-FM-143612-2 and 4; grooming claw: Ntg ¼ AMNH-FM-143612-3); Oc ¼ Omomys carteri (nail: Ocn ¼ UM 32258_a; grooming claw: Ocg ¼ DPC 25505); Tb, Teilhardina brandti (nail: Tbn ¼ USNM 540587; grooming claws: Tbc ¼ UF 334000; Tbd ¼ 409804; Tbe ¼ 411197); Tm ¼ Tetonius matthewi (grooming claw: UCMP 217999).

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

15

Table 7 Descriptive statistics of key variables for fossils and different taxonomic/phalanx-form groups. Negative values for FIA indicate a convex rather than concave proximal articular facet. Group

Teilhardina gc's Arapahovius gc Tetonius gc Omomys gc Notharctus gc Tarsius dp2 Tarsius dp3 Lemurid gc's Indriid gc's Cheirogaleid gc's Galagid gc's Falculae Erinaceid faculae Erinaceid gc's Ungulae

FSA

FIA

FTH ratio

n

Mean

SD

Range

n

Mean

SD

Range

n

Mean

SD

Range

3 1 1 1 1 12 12 17 10 11 9 38 3 7 77

61.50 55.60 59.80 48.90 62.20 63.14 56.53 63.03 66.93 56.53 56.49 88.28 71.57 69.48 82.31

3.27 e e e e 5.79 5.89 5.71 7.45 7.46 4.10 11.04 8.29 5.63 6.18

57.8e64.0 e e e e 53.4, 71.6 49.3, 64.2 52.6, 73.2 54.8, 79.7 41.4, 69.3 47.6, 61.1 62.0, 119.5 62.0, 76.5 62.0, 78.2 68.2, 97.0

3 1 1 1 1 12 12 17 10 11 9 38 3 7 77

51.10 40.32 26.53 72.66 19.20 27.33 31.01 7.58 6.32 37.23 36.12 122.53 102.98 113.43 64.20

15.55 e e e e 12.28 15.34 22.31 27.41 13.8 23.6 22.52 16.72 17.71 17.69

33.5e62.8 e e e e 4.4, 39.8 4.9, 53.8 44.3, 46.3 41.0, 42.9 17.6, 55.5 2.6, 84.6 88.0, 174.4 88.0, 121.0 96.2, 144.1 24.9, 104.7

3 1 1 1 1 12 12 17 10 11 9 38 3 7 77

0.34 0.32 0.35 0.36 0.22 0.29 0.24 0.20 0.22 0.14 0.22 0.59 0.46 0.51 0.20

0.04 e e e e 0.12 0.10 0.08 0.09 0.04 0.04 0.22 0.11 0.14 0.05

0.32e0.39 e e e e 0.15, 0.47 0.14, 0.45 0.13, 0.46 0.10, 0.36 0.06, 0.20 0.15, 0.27 0.22, 0.97 0.33, 0.53 0.37, 0.77 0.10, 0.34

Table 8 Ratio of second pedal distal phalanx to third or fourth pedal distal phalanx dimensions. Sample Teilhardina brandti large omomyid ungulae small omomyid ungulae large omomyid ungulae small omomyid ungulae large omomyid ungulae small omomyid ungulae Omomys carteri Notharctus tenebrosus Aotus Lemuridae Indriidae Cheirogaleiidae Galagidae Tarsiidae

gc specimen a

UF#’s UCMP#’sa UCMP 218344 UCMP 218000 UCMP 218344 UCMP 217999 UCMP 218344 DPC 25505 AMNH 143612 Multiple Multiple Multiple Multiple Multiple Multiple

Site

Age

Ratio 1

SE

Ratio 2

SE

N

WY140447 all all V74022 V74022 various V70214 UCM L93026 ALX-00-05 na na na na na na

Wa0 Wa3-5 Wa3-5 ~Wa5 ~Wa5 ~Wa3 ~Wa3 Br2 Br2 extant extant extant extant extant extant

0.813 0.960 na 0.964 na 0.954 na na 0.800 0.816 0.837 0.812 0.924 1.016 0.977

na na na na na na na na na e 0.025 0.021 0.045 0.060 0.019

0.620 0.686 0.674 0.692 0.672 0.677 0.689 0.802 0.600 0.566 0.601 0.578 0.663 0.710 0.699

na na na na na na na na na e 0.015 0.037 0.038 0.050 0.017

3 14 7 6 6 8 1 1 1 2 4 3 4 4 3

Abbreviations: dp ¼ distal phalanx; Ratio 1 ¼ grooming claw GM1 (dpL, dpW, dpH)/ungulae GM3 (dpL, dpW); Ratio 2 ¼ grooming claw GM2 (dpW, dpH)/ungulae GM3. See Table 2 legend for other abbreviations. See Table S8 for individual specimen data. ‘N’ refers to the number of ungulae averaged in the denominator of each comparison. a ‘UF#'s’ means the average of UF 334000, 409804, and 411197 were compared against the average value of ‘Wa0 omomyid ungulae’ which also constitutes 3 specimens (Table 3). ‘UCMP#’s’ means that the average values of 217999 and 218000 were compared against the average value of all the ‘large omomyid’ ungulae from Wa3 and Wa5.

average for the large ungular group (see Materials and methods for description of geometric mean calculations) was found to be 0.964 (Table 8; ratio 1). A reduced measurement ratio was also constructed (Table 8; ratio 2) so that we could compare UCMP 218344 and DPC 25505 (incomplete specimens from the Wasatch and Bridger Basins, respectively) to the ungular phalanges from the same sites. The reduced ratio represented size of the ‘grooming claw’ as the geometric mean of its proximal end dimensions only (GM3). In this case, extant taxa ranged from 0.57 to 0.71 (Table 8; ratio 2). UCMP 217999 and UCMP 218000 have a ratio of 0.686 to the large ungular phalanges. UCMP 218344 has a ratio of 0.674 to the small ungular phalanges. Computing these ratios for the Wa0 sample, including the T. brandti ‘grooming claw’ specimens and ungular phalanges from the same sites described by Rose et al. (2011), gives average values of 0.81 (ratio 1) and 0.62 (ratio 2). Computing ratio 2 for DPC 25505 against a sample of 11 omomyid ungular distal phalanges from Bridger Formation localities (Table 1) gives a value of 0.80 (Table 8). To further test if the grooming phalanges described in this study are of the correct size for omomyiform primates, species means of mandibular first molar dimensions of extant and fossil primates were regressed on measures of grooming phalanx size (Fig. 8). We recovered an isometric scaling relationship with a

strong correlation (R2 ¼ 0.88) for second pedal distal phalanx length vs. M1 length among extant primates (Fig. 8A). A second regression, including three extant erinaceids and measurements of the Eocene erinaceomorph Macrocranion, revealed a relationship with a shallower slope and larger intercept, such that the two regression lines occupy different plot areas for small distal phalanx lengths but converge for larger sizes (Fig. 8B). Importantly, at the small size of the focal fossils of this analysis (~2.2e3.2 mm in length), the 95% confidence intervals for prediction of tooth size from phalanx size are almost nonoverlapping. Significantly larger teeth are therefore predicted for an erinaceomorph with distal phalanges of 2.2e3.2 mm in length than for primates with phalanges of the same lengths. All of the included dental measures for Wasatch Fm. omomyid species fall within the 95% confidence intervals of the primate regression line when paired with recovered grooming phalanges, but mostly outside those of the eulipotyphlan region. Additionally, lipotyphlans known from the same deposits (Amphilemuridae indet., cf. Leipsanolestes sp., cf. Talpavoides sp., cf. Macrocranion sp.) also have tooth dimensions that fall within the confidence limits of the primate regression line, but entirely outside the limits of the insectivore regression line, suggesting the distal phalanx specimens are too big for these non-primate dentitions. A similar line of reasoning supports the assignment

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

16

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Figure 8. Regressions of log-transformed first mandibular molar size on phalanx size. Data for these plots are found in SOM Tables S8 and S9. There is a significant correlation between distal phalanx size and first mandibular molar size among primates and erinaceomorphs. However, erinaceomorphs have bigger teeth relative to their distal phalanges. Therefore, it is significant to note that most fossil primate taxa from the Wasatch Fm. localities have teeth in the size range expected for the fossil phalanges UCMP 217999, 218000, and 218344; only some of the primate teeth are in the range of expected sizes for the erinaceomorph regression. Therefore, the erinaceomorphs and other Wasatch Formation eulipotyphlans are too small to have had phalanges in the size range of those described here. Likewise, Teilhardina brandti teeth from the Willwood Fm. are in the size range predicted by grooming phalanges UF 334000, 411197, and 409804. Measurements for Macrocranian, an Eocene erinaceomorph come from Maier (1979). Abbreviations: PED ¼ proximal end dorsoventral depth; PEW ¼ proximal end mediolateral width.

of the Wa0 Bighorn Basin grooming phalanges to T. brandti, the only omomyiform known from the region during this interval. While the M1 lengths of T. brandti are somewhat high relative to lengths of the recovered grooming phalanges from the same sites

(Fig. 8B), the estimated proportion nevertheless falls into the 95% confidence interval for euprimates. The only erinaceomorphs contemporaneous with T. brandti (Macrocranion junnei, cf. Colpocherus sp., cf. Talpavoides sp.) have smaller M1 dimensions than

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

T. brandti, not larger, indicating their distal phalanges should be much shorter than those described here. 3.3. Cladistic results and character mapping A heuristic search in TNT using settings specified by Ni et al. (2013, 2016) explored 857,019,806,691 topologies. The search resulted in 57 most parsimonious trees with 13,647 steps. In terms of pivotal taxa and low level relationships, the tree is nearly identical to that published by Ni et al. (2016) and Boyer et al. (2017). Adding new character scorings or modifying previous scorings of characters of the grooming claws did not change the topology substantially. Optimization of character #1009 over the strict consensus tree suggests that grooming phalanges were lost from pedal dpII in the ancestral euarchontan (with a transition to falcular distal phalanges), based on the reconstruction of Erinaceus as the sister taxon to Euarchonta in this analysis (Fig. S2: node #215; but see discussion for a more likely scenario based on a more realistic outgroup sampling). The first transition relevant to this study occurred in the ancestral euprimate (Fig. S2: node #212), where grooming phalanges were gained (Fig. 9). The condition of pedal dpII becomes equivocal at node #235 (stem Anthropoidea) and may be either a grooming phalanx or ungular in form. However, the ancestral crown anthropoid (Fig. S2: node #230) is unequivocally reconstructed as having an ungular condition for its pedal dpII. Thus the re-evolution of grooming claws in Aotus and Callicebus appear as independent and unequivocal transitions in our study as well. 4. Discussion 4.1. Taxonomic attribution Three fossil distal phalanges from the Bighorn Basin (UF 409804, 334000, and 411197), another three from the Washakie Basin (UCMP 217999, 218000, and 218344), and one from the Bridger Basin (DPC 25505) are grooming phalanges assignable to omomyiforms based on: (1) their strong morphological similarity to the grooming phalanges of extant strepsirrhines and/or tarsiers, and (2) an absolute size that is proportionally correct for the omomyiform teeth and ungular phalanges from the sampled localities. Beyond an attribution to omomyiforms, we tentatively assign the fossils to specific taxa that are appropriate for their size. The Bighorn Basin specimens are straightforward to attribute: they are all from the Wa0 faunal zone and of a size appropriate for the ungular phalanges (Table 8) and dentition (Fig. 8) of T. brandti. This case is further simplified by the fact that T. brandti is the only omomyiform known from this faunal zone and that the specimens are much smaller than would be expected for those of the Wa0 adapiform C. torresi. From the Washakie Basin, we assign UCMP 217999 to Tetonius matthewi, based on the size (Fig. 8) and the presence of teeth of this species at the same site, which is Wa3 in age. We assign UCMP 218344 to Tetonoides pearcei, due to its smaller size, although it is also Wa3 in age and T. matthewi also occurs at the locality yielding this specimen. UCMP 218000 is similar in size to UCMP 217999, but from a younger, Wa5 locality, where T. matthewi is absent. Therefore, we assign UCMP 218000 to Arapahovius gazini, which is present in Wa5 and has a dental size similar to that of T. matthewi. Finally, there can be little doubt that the grooming phalanx from Omomys quarry (DPC 25505) belongs to a species of Omomys. It is too small to belong to a notharctid (Br2-3 species of Smilodectes and Notharctus are similar in body size to one another and have distal phalanges roughly twice as large as DPC 25505; Fig. 2; Table 3). DPC 25505 is slightly larger than would be expected based on ungular

17

phalanges and dentitions attributed to Omomys (Fig. 8; Table 8). However, this may mostly reflect the exceptionally large dorsoventral depth of the proximal end, with its pronounced flexor tubercle. While Hemiacodon often co-occurs with Omomys during the Bridgerian and is thought to be slightly larger, there is apparently no evidence for it at Omomys Quarry. Meanwhile, the sample of Omomys from this site is comprised of over 200 specimens (Murphey et al., 2001). 4.2. Morphological indications of a grooming claw The morphometric analysis based on linear and angular measurements presented here suggests that all extant primate grooming phalanges are similar to each other, and distinct from the grooming phalanges of non-primates. The finding that primate grooming phalanges are morphologically distinct is consistent with previous analyses of this trait (Maiolino et al., 2011, 2012; von Koenigswald et al., 2012). The automated morphometric analysis results are consistent with measurement analysis results because they distinguish falcular, ungular, and grooming phalanges. However, automated results also suggest that tarsier grooming phalanges from digit II and III are distinct from those of other euprimates. This result is consistent with the interpretation of Ni et al. (2013, 2016), as well as the discussion of Maiolino et al. (2011) and Gebo et al. (2017), who described qualitative traits distinguishing grooming phalanges of tarsiers from those of all strepsirrhines. While we have not quantitatively verified many of the traits they described, tarsiers do tend to have a more pronounced flexor tubercle on their grooming phalanges (but not on their more lateral distal phalanges) compared to those of extant strepsirrhines (Table 7). While the automated approach suggests that erinaceid grooming phalanges have a shape similar to that of primate grooming phalanges (Fig. 6), and therefore that primate and non-primate grooming claws might be difficult to distinguish in practice, this outcome fails to capture the differences in included angle of the articular facet (FIA; Fig. 7, Table 7, SOM Table S4) between the two groups of grooming phalanges. Instead, this aspect of the automated result appears to be driven by both groups expressing proximodistally long, mediolaterally narrow, and dorsoplantarly shallow shafts. Other distal phalanx types tend to have relatively more expansion in either their dorsoplantar axis (falcular) or their mediolateral axis (ungular). These outcomes have several implications. First, according to the results from both sets of morphometric analyses (Figs. 5 and 6) leave little ambiguity that they are best attributed to euprimates. Specifically, according to the measurement analyses (Fig. 5), the fossils are similar to the grooming phalanges of other euprimates and distinguished from non-primate ‘grooming phalanges’ in having a shaft with more pronounced dorsal canting and an articular facet that tends to be less concave. An additional qualitative feature observed in DPC 25505 that appears to strengthen its attribution to euprimates (specifically Omomys) is the pronged proximal margin of the apical tuft: these ‘prongs’ (aka ungual spines) contribute to the ‘arrow head-like’ appearance of the complete apical tuft in ungula of various ‘prosimians,’ including Omomys (Figs. 2 and 3). The automated morphometric results, while also revealing the fossils to be closest to euprimates, invite interpretations with implications for the evolutionary processes leading to the taxonomic distribution of ‘grooming claws’ in euprimates. One potential interpretation of the plot (Fig. 6) supports the argument that grooming phalanges evolved multiple times in euprimates (von Koenigswald et al., 2012; Ni et al., 2013), due to tarsier and strepsirrhine grooming phalanges being distinguished by their PC scores. An additional possible interpretation of the plot (Fig. 6) is that tarsiers and T. brandti

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

18

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Figure 9. Summary of phylogenetic results and character optimization. This figure depicts a summary of the strict consensus cladogram resulting after alteration of states of characters #1009-1010 (relating to ‘grooming claws’) of and reanalysis of matrix from Ni et al. (2016). Three clades represented by gray background are (from top to bottom) Adapiformes, Omomyiformes/Tarsiiformes, and Anthropoidea. Bold taxa are those where grooming phalanx morphology is observable. No omomyid species were known for this morphology before this study. Curved arrows leading from nodes with black dots indicate reconstructed transformations in the form of pedal digit dpII (ch #1009). Nodes: 1 e transformation from falcula on pedal dpII (e.g, Box B e Tupaia, Carpolestes) to a grooming claw (e.g., Box C e Teilhardina, Tarsius, Notharctus, Lemur); 2 e transformation from grooming claw (Box C) on pedal dpII to ungular phalanx (e.g., Box D e Chlorocebus, Alouatta); 3 e transformation from ungular phalanx (Box D) to grooming claw in Callicebus (Box E); 4 e transformation from ungular phalanx (Box D) to grooming claw in Aotus (Box E); 5 e transformation from ungular phalanx (Box D) to tegulae in Callitrichinae (e.g., Box F e Callithrix). Box A shows the grooming phalanx morphology found in species of Erinaceus; we do not consider this to reflect the primitive boreoeutherian condition, though our character optimization did in fact reconstruct it as such.

have a more primitive form of grooming phalanx because their grooming phalanges are more similar to falcular phalanges than are those of other omomyiforms (Arapahovius and Tetonius) and strepsirrhines (including adapiforms).

The first interpretation of the automated results (convergent acquisition of grooming phalanges by tarsiers and strepsirrhines) is refuted by our character optimization suggesting that grooming claws in these groups were inherited from the euprimate last

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

common ancestor. On the other hand, the second interpretation of tarsier and T. brandti grooming phalanges as primitive is consistent with our character optimization and is further supported by the observation that all omomyid grooming phalanges and at least some tarsier species have a more falcula-like flexor tubercle, as discussed above. If the positions of T. brandti and Tarsius in the automated morphometric plots are to be taken as indicative of their ‘primitiveness,’ then the position of other omomyiform fossils close to extant strepsirrhines indicates subsequent parallel evolution of the grooming phalanx, as also suggested by Maiolino et al. (2012). Additional evidence for parallel evolution of the ‘grooming claw’ can be found in qualitative features of the Omomys ‘grooming claw’ that make it similar to that of Notharctus and extant strepsirrhines, as discussed below. 4.3. Qualitative characters of grooming phalanges As alluded to earlier, one qualitative feature that aids in evaluating ‘grooming claw’ evolution is the size and morphology of the grooming phalanx basal nutrient foramen. Like erinaceids, both omomyiforms and adapiforms retain a large nutrient foramen at the base of their grooming phalanges. The fact that a strong basal nutrient foramen is also prevalent in falcular phalanges, but not ungular phalanges may suggest their presence in fossil grooming phalanges is a primitive retention from an earlier, falcular stage for this digit. As an additional similarity to falcular phalanges, the nutrient foramen in adapiform (but not omomyiform) grooming phalanges has a large communication with the medullary cavity of the shaft. Thus, whereas the overall morphology of Bridgerian adapiform grooming phalanges appears more derived than that of the largely Wasatchian sample of omomyiforms identified here (Figs. 2 and 3), the configuration of the basal nutrient foramen is an exception because it appears more primitive in adapiform grooming phalanges than in those of omomyiforms. The basal nutrient foramen of the omomyiform grooming phalanges appears to have only a weak communication with the medullary cavity of the shaft, suggesting it has less of a functional role in providing neurovasculature to the medullary cavity. Loss of functionality of the basal nutrient foramen seems a likely precursor to its complete loss in extant primate grooming phalanges (Fig. 3). This difference between adapiform and omomyiform grooming phalanges is paralleled by differences in the ungular morphology of the two clades: the basal nutrient foramen in omomyiform ungular phalanges tends to be reduced or absent, while that in adapiforms tends to be larger and to communicate with the medullary cavity (Maiolino et al., 2012). We find it interesting that omomyiform grooming phalanges appear more falcula-like in their flexor tubercle projection, while adapiform grooming phalanges are more primitive in the gauge of the canal connecting the basal nutrient foramen to the shaft medullary cavity. We consider this a reflection of mosaic evolution. The presence of mosaicism in the primate grooming phalanges described here suggests that the ancestral euprimate was characterized by a grooming phalanx with an even greater number of primitive traits. Under this scenario, the ancestral euprimate grooming phalanx could give rise to a ‘Teilhardina-like’ shape, losing the strong nutrient foramen connection to the medullary cavity early on, or alternatively an ‘adapiform-like’ configuration that loses the flexor tubercle early on. Few fossil primate species are hypothesized to be more basal than T. brandti. Potential candidate taxa include the basal adapiform D. provincialis (see Boyer et al., 2017) and the omomyiform A. achilles (see Ni et al., 2013). If D. provincialis and A. achilles are indeed more primitive, then their

19

second pedal distal phalanges may have both a strong flexor tubercle and a large canal connecting the basal nutrient foramen to the medullary cavity. Finally, the grooming phalanx of Omomys appears more derived than the grooming claws of the other, older omomyiforms. Though DPC 25505 retains the prominent flexor tubercle seen in tarsiers and other omomyiforms, its grooming phalanx looks more strepsirrhine-like in being mediolaterally broader and in having an apical tuft with sharp, proximally projecting ‘prongs’ (a.k.a. ‘ungual spines’) at its proximal base (Fig. 2). 4.4. Number of primate digits bearing grooming claws/phalanges Given our interpretation of the tarsier grooming phalanx as primitive relative to those of extant strepsirrhines, it is reasonable to question whether the number of ‘grooming claws’ in tarsiers is also primitive. That is, tarsiers actually have two ‘grooming claws’ (on the dpII and dpIII), unlike strepsirrhines. We ask whether having two ‘grooming claws’ is also a primitive retention. The only way to answer this question definitively would be with articulated feet of basal euprimate species. The only basal haplorhine with articulated feet is A. achilles, which retains only an impression of the dpII, but still preserves the dpIII (Ni et al., 2013). However, the condition of dpIII in A. achilles is ambiguous as it appears to be crushed against the intermediate phalanx and no mediolateral view is available. Looking to potential fossil strepsirrhines, von Koenigswald et al. (2012) identified E. koenigswaldi as having a single grooming phalanx on its articulated foot (on dpII). However, von Koenigswald et al. (2012) identified both the second and third pedal distal phalanges of N. tenebrosus as grooming phalanges. In contrast, the analyses presented here and those of Maiolino et al. (2012) are inconsistent with a grooming phalanx designation for dpIII in Notharctus. This raises the question of what condition is expressed by more basal adapiforms, primarily species of Cantius and Donrussellia; unfortunately, no skeletons of these taxa have yet been described. 4.5. Variation in second digit distal phalanges: developmental correlation? Omomyiform grooming phalanges appear to show a trend of becoming more ‘ungular’ from the early Wasatchian (~56 Ma) to the late Bridgerian (~47 Ma), and contemporaneous Bridgerian adapiform grooming claws (Notharctus, e.g., AMNH 143612-3) appear even more ungular in dorsal view (Fig. 2C). There is evidence that the morphologies of adjacent postaxial digits tend to be correlated genetically and developmentally in consistent ways (Kavanagh et al., 2013). Therefore, we should be wary of interpreting all of the evolutionary trends in dpII as a result of direct selection without considering the potential effects of selection on the other digits or non-selective modes of evolution. One specific question that may be asked is whether the process causing primate dpII grooming phalanges to become more ‘ungular’ over time was direct selection for prehensile functions of pedal digit II, or whether the process was mainly a result of indirect selection on adjacent postaxial phalanges. If selection for prehensile grasping was acting on digits IIIeV, as is frequently postulated to be the case during early euprimate evolution (Gebo, 1985, 2004; Sargis et al., 2007), then this process could conceivably cause a correlated response in digit II. Under this scenario, the development of ‘ungular’ morphology is predicted in dpII even if it was not experiencing direct selection for grasping or experienced contrasting selection for ‘grooming functions.’

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

20

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

4.6. Adaptive value of grooming claws One may also question the premise that features defining a grooming phalanx represent a classic, Darwinian selective adaptation (Williams, 1966). It is possible that these features arose as a result of non-selective evolution (i.e., drift) and/or as a consequence of selection acting elsewhere, having been later co-opted for a grooming function (Gould and Lewontin, 1979; Gould and Vrba, 1982). The observations of this study contribute evidence for a direct, adaptive value of grooming claws rather than for their emergence as spandrels and later use in grooming behaviors as exaptations. It is thought that novel adaptive trends often appear when a structure with a formerly critical function becomes redundant with other structures or when selection is relaxed for other reasons. The evolution of a ‘double jaw joint’ in mammaliamorphs is a prime example. Once the dentary and squamosal bones invaded the craniomandibular joint, the articular and quadrate were no longer committed to selection for maintaining jaw function. Drift and eventually selection for other functions could transform the latter bones into the malleus and incus to form the mammalian triossicular middle ear along with the stapes (Crompton, 1962; Meng et al., 2018). Gebo (1985, 1988) has made the argument that the primitive morphofunctional mode of euprimate grasping is what he calls the ‘IeV grasp’ mode and suggested that it was present in notharctine adapiforms such as species of Cantius and Notharctus. In this mode, the hallux and outer digits are critical for maintaining the grasp, but the inner digits (IIeIII) are less important. There are a number of osteological features that apparently reflect this grasp mode in the tarsals, metatarsals and digits. With regard to the digits, an accentuated ectaxonic digit length pattern appears to reflect a IeV grasp, since digits III and especially II become reduced in length and robusticity, while the outer digits (IVeV) increase (Maiolino et al., 2012:Fig. 16). Perhaps grooming claws were allowed to evolve due to relaxed selection for clinging in the IeV grasp mode. This idea predicts that ectaxony and IeV grasping characterized the earliest euprimates. Potential stem primate plesiadapiforms Carpolestes simpsoni (Bloch and Boyer, 2002) and Plesiadapis cookei (Boyer and Gingerich, in press) exhibit pronounced metatarsal ectaxony. The most basal known haplorhine, A. achilles also exhibits pronounced ectaxony in both its metatarsal and digit length patterns (Ni et al., 2013). While neither grasp pattern nor axony can be confirmed in T. brandti, this species is exceedingly similar to Archicebus in most respects (Boyer et al., 2013; Ni et al., 2013; Dagosto et al., 2017), so it is very likely to have had a IeV grasp and strong ectaxony. Therefore, the presence of grooming phalanges in T. brandti and species of Notharctus indicates that these structures arose initially in the context of a IeV grasp pattern where selective pressures for grasping functions were likely reduced on digits IIeIII. In certain lorises, manual ectaxony is so extreme that digit II is reduced to a stump with no claw or nail at all. Rather than becoming vestigial, another potential pathway for a digit II with no grasping function is that it can be selected for other functions. If there is reduced functional demand on the second digit for grasping in ectaxonic feet, the morphology present in grooming phalanges such as the dorsally projecting shaft and proximally positioned flexor tendon insertion could represent an adaptation simply to ‘get the claw out of the way’ during locomotion and grasping. In this case, use of the dpII in grooming would be an exaptation. However, we think there is good evidence against this interpretation: (1) lemurs with a IeII grasp (i.e., indriids, lemurids, and Lepilemur; Gebo, 1985) retain a ‘grooming claw’ and use it in autogrooming behaviors, and (2) anthropoids have re-evolved the

‘grooming claw’ but lack an ectaxonic, strepsirrhine-like IeV grasp complex that would benefit from ‘getting the dpII tip out of the way.’ Grooming behaviors themselves therefore represent the likely candidates for producing the selective pressures that molded ‘grooming claw’ morphology. While grooming behavior plays a ubiquitous central role in sociality of all extant primate species, the morphology of the dpII in adapiforms and omomyiforms represents the only fossil evidence for specialized anatomy devoted to grooming until the earliest record of toothcombs, some 15 million years later (Seiffert et al., 2003). If ‘grooming claws’ have been under selection for grooming behaviors since the beginning of a more ectaxonic foot, then it seems likely that the pattern observed here, whereby grooming claws of early euprimates become increasing similar to ungulae over time, is a result of developmental correlation (Kavanagh et al., 2013) rather than selection for a grasping function. Indeed, the ways in which even the earliest primate grooming phalanges resemble ungular phalanges may be a result of developmental correlation rather than selection for prehensile grasping. The loss of grooming claws in anthropoids may correlate with reduced importance of autogrooming due to increasing social complexity in anthropoids relative to prosimians (Dunbar, 1998). Reducing the importance of autogrooming selection on dpII form would allow selection for grasping to take over in anthropoids with a IeII pedal grasp pattern. This is also consistent with the hypothesis that Aotus and Callicebus have re-evolved grooming claws, because they keep smaller social groups and may have fewer opportunities for allogrooming. 4.7. Potential scenario for evolutionary transitions of pedal distal phalanges The results of this study add to our understanding of ‘grooming claw’ morphology, antiquity, and taxonomic distribution among primates. Given these findings, we propose the following narrative for the evolution of primate ‘grooming claws’, which is intended as a set of falsifiable hypotheses to be tested through further fossil discovery and analysis (Fig. 9). Our descriptions of morphological transitions in pedal dpII form are based on the results of character optimization (SOM Fig. S2) for the most part. One result shown in SOM Figure 2, which we do not embrace (Fig. 9), is the reconstructed primitive presence of a ‘grooming claw’ in noneuarchontans. This is largely an artifact of insufficient taxon sampling relative to the goal of estimating the distribution of ‘grooming claws’ outside Euarchonta. In other words, although some species of Erinaceidae, our outgroup, have a ‘grooming claw’ on pedal dpII, we think that the more general condition for primitive placental mammals (including non-euprimate euarchontans) is to have a falcular distal phalanx on pedal dpII. This conclusion implies that erinaceid grooming claws are derived within Laurasiatheria. To be clear, we propose that basal members of most placental clades (including stem primates) lacked ‘grooming claws’ on pedal dpIIedpIII and had falcula-bearing phalanges like plesiadapiforms (Figs. 3 and 9: Carpolestes; Bloch and Boyer, 2002) and treeshrews. In arboreal non-euprimate euarchontans, all lateral digits were relied on for claw clinging. As the hallucal digit became more developed in stem primates (potentially, but not necessarily including plesiadapiforms) for opposed IeV grasping, digits IIeIII became less important for maintaining purchase on the substrate and were freed to selection for other functions. As a result of selection on digit II (and possibly digit III) for grooming functions, the last common ancestor of crown primates already possessed grooming phalanges similar to those seen in tarsiers and T. brandti (the most primitive primate for which grooming claws are known).

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

These grooming phalanges retained a large degree of similarity to the falcular phalanges of their predecessors (Figs. 3, 6, and 9). Furthermore, the ancestral primate grooming phalanx retained a large, basal nutrient foramen with a strong medullary communication, like that of falcular phalanges and grooming phalanges known for adapiforms and erinaceids. This primitive form was then modified in parallel in the strepsirrhine lineage and at least some omomyiform lineages to share fewer similarities with falculae and more with ungulae, as reflected by the automated morphometric analysis (Fig. 6). Such parallelism was probably a result of developmental correlation with lateral digits IV and V (Kavanagh et al., 2013). Loss of a large, functional nutrient foramen occurred more basally in haplorhines than in strepsirrhines. The ‘grooming claw’ was lost altogether in basal anthropoids as a result of using more horizontal substrates, more adducted foot postures, loss of a IeV grasp pattern, and increased social complexity where autogrooming was less important. It reappeared in Aotus and Callicebus as a result of decreased group size and correspondingly lessened opportunities for allogrooming. The grooming phalanges of extant strepsirrhines, Notharctus, and Omomys share greater morphological similarities with each other than with those of tarsiers, Teilhardina, and erinaceids because they have experienced more prolonged indirect selection for a more ungula-like morphology. Grooming phalanges of anthropoids have a more ‘ungular’ appearance because, unlike the other taxa sporting ‘grooming claws,’ anthropoid ‘grooming claws’ were actually derived from ungulae (much like falcula-like tegulae of Daubentonia and callitrichids were derived from ungulae; Fig. 9). 5. Summary and conclusions Newly described specimens and comparative analyses provide the first compelling evidence that members of the omomyiforms, one of the two early Eocene euprimate radiations, possessed ‘grooming claws’ on at least the second pedal digit. Second pedal digit grooming claws are now documented in all major extant euprimate groups except for catarrhines, and with the evidence provided by the fossil record, they appear to be ubiquitous in strepsirrhines and non-anthropoid haplorhines. This wide distribution of grooming phalanges across the primate tree and their occurrence among the oldest taxa attributed to crown primates (T. brandti, ~56 Ma) add support to the hypothesis that a ‘grooming claw’ was present in the ancestral euprimate. Additionally, our results suggest that tarsiids and basal omomyiforms have a more primitive ‘grooming claw’ form than other euprimates, while adapiforms have a more primitive nutrient foramen configuration. This mosaicism suggests that the ancestral euprimate had a ‘grooming claw’ like that of tarsiers and a nutrient foramen like that of adapiforms. It may be that the ancestral euprimate had at least two grooming claws, and that the digit three ‘grooming claw’ was lost in parallel in haplorhines and strepsirrhines, although current fossil evidence in support of this hypothesis is limited. Further tests of our conclusions will require more postcranial information on other omomyiforms and adapiforms. In particular, articulated material of basal euprimates will be critical. Understanding the evolutionary history of 'grooming claws' will eventually contribute to a more complete picture of the morphological changes characterizing euprimate origins and aid in understanding the selective pressures that characterized this evolutionary transition as well as other transitions in early euprimate evolution. Acknowledgements We thank G.F. Gunnell and C. Riddle of the Duke Lemur Center Division of Fossil Primates, and N. Simmons, E. Westwig and N.

21

Duncan of the American Museum of Natural History for approving and facilitating loans of comparative material. We thank J. Thostenson for scanning specimens. NSF grants to D.M.B. and E.R. Seiffert (BCS-1317525), to D.M.B. and G.F. Gunnell (BCS-1440742), to D.M.B., G.F. Gunnell and R.F. Kay (DBI-1458192), to J.I.B. (BCS1440558), D.M.B. (BCS-1552848), S.A.M. (BCS 1341075), and to J.I.B., R. Secord, and J.S. Krigbaum (EAR-0640076) supported this work. A Leakey Foundation grant supported S.A.M. Finally, an IMLS grant to P.A.H., C.R. Marshall, and L.D. White (MA-30-15-0336) supported processing and curation of Wasatch Fm. collections. Fossils were collected under Bureau of Land Management permits to J.I.B. (PA04WY-113 and PA10-WY-185) and G.F. Gunnell (157-WY-PA94). This is Duke Lemur Center publication #1395. This is University of Florida Contribution to Paleobiology 851. Supplementary Online Material Supplementary online material related to this article can be found at https://doi.org/10.1016/j.jhevol.2018.03.010. References 3D Systems Inc, 2013. Geomagic Studio. 3D Systems Inc, Rock Hill. Anemone, R.L., Covert, H.H., 2000. New skeletal remains of Omomys (Primates, Omomyidae): functional morphology of the hindlimb and locomotor behavior of a Middle Eocene primate. Journal of Human Evolution 38, 607e633. Baczynski, A.A., McInerney, F.A., Wing, S.L., Kraus, M.J., Bloch, J.I., Boyer, D.M., Secord, R., Morse, P.E., Fricke, H.C., 2013. Chemostratigraphic implications of spatial variation in the Paleocene-Eocene Thermal Maximum carbon isotope excursion, SE Bighorn Basin, Wyoming. Geochemistry, Geophysics, Geosystems 14, 4133e4152. Beard, K.C., Dagosto, M., Gebo, D.L., Godinot, M., 1988. Interrelationships among primate higher taxa. Nature 331, 712e714. Bloch, J.I., Boyer, D.M., 2002. Grasping primate origins. Science 298, 1606e1610. Bloch, J.I., Silcox, M.T., Boyer, D.M., Sargis, E.J., 2007. New Paleocene skeletons and the relationship of plesiadapiforms to crown-clade primates. Proceedings of the National Academy of Sciences USA 104, 1159e1164. Bourque, J.R., Howard Hutchison, J., Holroyd, P.A., Bloch, J.I., 2015. A new dermatemydid (Testudines, Kinosternoidea) from the Paleocene-Eocene Thermal Maximum, Willwood Formation, southeastern Bighorn Basin, Wyoming. Journal of Vertebrate Paleontology 35, e905481. Boyer D.M., Gingerich P.D., Skeleton of late Paleocene Plesiadapis cookei (Mammalia, Euarchonta, Plesiadapiformes): life history, locomotion, and phylogenetic relationships, 2018, University of Michigan Papers on Paleontology, in press. Boyer, D.M., Costeur, L., Lipman, Y., 2012. Earliest record of Platychoerops (Primates, Plesiadapidae), a new species from Mouras Quarry, Mont de Berru, France. American Journal of Physical Anthropology 149, 329e346. Boyer, D.M., Lipman, Y., St Clair, E., Puente, J., Patel, B.A., Funkhouser, T.A., Jernvall, J., Daubechies, I., 2011. Algorithms to automatically quantify the geometric similarity of anatomical surfaces. Proceedings of the National Academy of Sciences USA 108, 18221e18226. Boyer, D.M., Puente, J., Gladman, J.T., Glynn, C., Mukherjee, S., Yapuncich, G.S., Daubechies, I., 2015. A new fully automated approach for aligning and comparing shapes. The Anatomical Record 298, 249e276. Boyer, D.M., Seiffert, E.R., Gladman, J.T., Bloch, J.I., 2013. Evolution and allometry of calcaneal elongation in living and extinct primates. PLoS One 8, e67792. Boyer, D.M., Gunnell, G.F., Kaufman, S., McGeary, T., 2016. MorphoSource: Archiving and sharing digital specimen data. The Paleontological Society Papers 22, 157e181. Boyer, D.M., Toussaint, S., Godinot, M., 2017. Postcrania of the most primitive euprimate and implications for primate origins. Journal of Human Evolution 111, 202e215. Cartmill, M., 1974. Pads and claws in arboreal locomotion. In: Jenkins, F.A. (Ed.), Primate Locomotion. Academic Press, New York, pp. 45e83. Cates, J., Fletcher, P.T., Styner, M., Hazlett, H.C., Whitaker, R., 2008. Particle-based shape
kely, G. analysis of multi-object complexes. In: Metaxas, D., Axel, L., Fichtinger, G., Sze (Eds.), International Conference on Medical Image Computing and ComputerAssisted Intervention e MICCAI 2008. Springer-Verlag, Berlin, pp. 477e485. Chester, S.G.B., Bloch, J.I., Secord, R., Boyer, D.M., 2010. A new small bodied species of Palaeonictis (Creodonta, Oxyaenidae) from the Paleocene-Eocene thermal maximum. Journal of Mammalian Evolution 17, 227e243. Covert, H.H., Hamrick, M.W., 1993. Description of new skeletal remains of the early Eocene anaptomorphine primate Absarokius (Omomyidae) and a discussion about its adaptive profile. Journal of Human Evolution 25, 351e362. Crompton, A.W., 1962. The evolution of the mammalian jaw. Evolution 14, 431e439. Cuozzo, F., 2002. Using extant patterns of dental variation to identify species in the primate fossil record: a case study of middle Eocene Omomys from the Bridger Basin, southwestern Wyoming. Primates 49, 101e115.

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010

22

D.M. Boyer et al. / Journal of Human Evolution xxx (2018) 1e22

Dagosto, M., 1988. Implications of postcranial evidence for the origin of euprimates. Journal of Human Evolution 17, 35e56. Dagosto, M., Gebo, D., Ni, X., Smith, T., 2017. Estimating body size in early primates: the case of Archicebus and Teilhardina. Journal of Human Evolution 115, 8e19. Dagosto, M., Gebo, D.L., Beard, K.C., 1999. Revision of the Wind River faunas, Early Eocene of central Wyoming. Part 14. Postcranium of Shoshonius cooperi (Mammalia : Primates). Annals of Carnegie Museum 68, 175e211. Dunbar, R., 1998. The social brain hypothesis. Brain 9, 178e190. Dunn, R.H., Sybalsky, J.M., Conroy, G.C., Rasmussen, D.T., 2006. Hindlimb adaptations in Ourayia and Chipetaia, relatively large-bodied omomyine primates from the middle Eocene of Utah. American Journal of Physical Anthroplogy 131, 303e310. FEI Visualization Science Group, 2009. AVIZO, Version 8.1. Mercury Computer Systems, Burlington. Franzen, J.L., Gingerich, P.D., Habersetzer, J., Hurum, J.H., von Koenigswald, W., Smith, B.H., 2009. Complete primate skeleton from the middle Eocene of Messel in Germany: morphology and paleobiology. PLoS One 4, e5723. Gebo, D.L., 1985. The nature of the primate grasping foot. American Journal of Physical Anthropology 67, 269e277. Gebo, D.L., 1988. Foot morphology and locomotor adaptation in Eocene primates. Folia Primatologica 50, 3e41. Gebo, D.L., 2004. A shrew-sized origin for Primates. Yearbook of Physical Anthropology 47, 40e62. Gebo, D.L., Beard, K.C., Ni, X., Dagosto, M., 2015. Distal phalanges of Eosimias and Hoanghonius. Journal of Human Evolution 86, 92e98. Gebo, D.L., Dagosto, M., Ni, X., Beard, K.C., 2017. Phalangeal morphology of Shanghuang fossil primates. Journal of Human Evolution 113, 38e82. Gebo, D.L., Smith, T., Dagosto, M., 2012. New postcranial elements for the earliest Eocene fossil primate Teilhardina belgica. Journal of Human Evolution 65, 305e218. Gingerich, P.D., 2012. Primates in the Eocene. Palaeobiodiversity and Palaeoenvironments 92, 649e663. Godinot, M., 1992. Early euprimate hands in evolutionary perspective. Journal of Human Evolution 22, 267e283. Goloboff, P., Catalano, S., 2016. TNT version 1.5, including a full implementation of phylogenetic morphometrics. Cladistics 32, 221e238. Goloboff, P., Farris, J., Nixon, K., 2003. TNT: Tree analysis using new technology, version 1.0. Program and documentation. Goloboff, P., Farris, J., Nixon, K., 2008. TNT, a free program for phylogenetic analysis. Cladistics 24, 1e13. Gould, S.J., Lewontin, R.C., 1979. The spandrels of San Marco and the Panglossian pradigm: a critique of the adaptationist programme. Proceedings of the Royal Society of London B 205, 581e598. Gould, S.J., Vrba, E.S., 1982. Exaptationda missing term in the science of form. Paleobiology 8, 4e15. Gregory, W.K., 1920. On the structure and relations of Notharctus, an American Eocene primate. Memoires of the American Museum of Natural History 3, 45e243. Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: Paleontological statistics software package for education and data analysis. Palaeontologia Electronica 4, 4. Hamrick, M.W., 1998. Functional and adaptive significance of primate pads and claws: Evidence from New World anthropoids. American Journal of Physical Anthropology 106, 113e127. Hamrick, M.W., 1999. First carpals of the Eocene primates family Omomyidae. Contributions from the Museum of Paleontology. University of Michigan 30, pp. 191e198. Hamrick, M.W., Alexander, J.P., 1996. The hand skeleton of Notharctus tenebrosus (Primates, Notharctidae) and its significance for the origin of the primate hand. American Museum Novitates 3182, 1e20. Jones, F.W., 1916. Arboreal Man. Arnold, London. Kavanagh, K.D., Shoval, O., Winslow, B.B., Alon, U., Leary, B.P., Kan, A., Tabin, C.J., 2013. Developmental bias in the evolution of phalanges. Proceedings of the National Academy of Sciences USA 110, 18190e18195. Kim, V.G., Li, W., Mitra, N.J., DiVerdi, S., Funkhouser, T., 2012. Exploring collections of 3D models using fuzzy correspondences. ACM Transactions on Graphics 31, 54. Le Gros Clark, W.E., 1971. The Antecedents of Man, 3rd ed. Quadrangle/New York Times Book Co., New York. Maier, W.,1979. Macrocranion tupaiodon, an adapisoricid (?) insectivore from the Eocene €ontologische Zeitschrift 53, 38e62. of ‘Grube Messel’ (Western Germany). Pala Maiolino, S.A., 2015. Comparative morphology of primate distal phalanges: implications for early primate evolution and the origins of the primate nail. Ph.D. Dissertation. Stony Brook University. Maiolino, S., Boyer, D.M., Rosenberger, A., 2011. Morphological correlates of the grooming claw in distal phalanges of platyrrhines and other primates: a preliminary study. Anatomical Record 294, 1975e1990. Maiolino, S., Boyer, D.M., Bloch, J.I., Gilbert, C.S., Groenke, J., 2012. Evidence for a grooming claw in a North American adapiform primate: implications for anthropoid origins. PLoS One 7, e29136. Meng, J., Bi, S., Zheng, X., Wang, X., 2018. Ear ossicle morphology of the Jurassic euharamiyidan Arboroharamiya and evolution of mammalian middle ear. Journal of Morphology 279, 441e457.

Murphey, P.C., Torick, L.L., Bray, E.S., Chandler, R., Evanoff, E., 2001. Taphonomy, fauna, and depositional environment of the Omomys quarry, an unusual accumulation from the Bridger Formation (Middle Eocene) of southwestern Wyoming (USA). In: Gunnell, G.F. (Ed.), Eocene Biodiversity: Unusual Occurrences and Rarely Sampled Habitats. Kluwer Academic/Plenum Press, New York, pp. 361e401. Murphey, P.C., Townsend, K.B., Friscia, A.R., Westgate, J., Evanoff, E., Gunnell, G.F., 2017. Paleontology and stratigraphy of Middle Eocene rock units in the southern Green River and Uinta Basins, Wyoming and Utah. Geology of the Intermountain West 4, 1e53. Napier, J.R., 1961. Prehensility and opposability in the hands of primates. Symposia of the Zoological Society of London 5, 115e132. Ni, X., Gebo, D., Dagosto, M., Meng, J., Tafforeau, P., Flynn, J.J., Beard, K.C., 2013. The oldest known primate skeleton and early haplorhine evolution. Nature 498, 60e64. Ni, X., Li, Q., Li, L., Beard, K.C., 2016. Oligocene primates from China reveal divergence between African and Asian primate evolution. Science 352, 673e677. O'Higgins, P., Jones, N.G.B., 1998. Facial growth in Cercocebus torquatus: an application of three-dimensional geometric morphometric techniques to the study of morphological variation. Journal of Anatomy 193, 251e272. Paniagua, B., Bompard, L., Cates, J., Whitaker, R., Datar, M., Vachet, C., Styner, M., 2012. Combined SPHARM-PDM and entropy-based particle systems shape analysis framework. In: Proceedings SPIE 8317, Medical Imaging 2012. Biomedical Applications in Molecular, Structural, and Functional Imaging 83170L. https://doi.org/10.1117/12.911228. Patel, B.A., Maiolino, S.A., 2016. Morphological diversity in the digital rays of primate hands. In: Kivell, L., Lemelin, P., Schmitt, D. (Eds.), The Evolution of the Primate Hand: Anatomical, Developmental, Functional, and Paleontological Evidence. Springer, New York, pp. 55e100. R Core Team, 2014. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. Rose, K.D., Chester, S.G.B., Dunn, R.H., Boyer, D.M., Chew, A.E., Bloch, J.I., 2011. New fossils of the oldest North American euprimate Teilhardina brandti (Omomyidae) from the Paleocene-Eocene Thermal Maximum. American Journal of Physical Anthropology 146, 281e305. Sargis, E.J., Boyer, D.M., Bloch, J.I., Silcox, M.T., 2007. Evolution of pedal grasping in Primates. Journal of Human Evolution 53, 103e107. Savage, D.E., Waters, B.T., 1978. A new omomyid primate from the Wasatch Formation of southern Wyoming. Folia Primatologica 30, 1e29. Secord, R., Bloch, J.I., Chester, S.G.B., Boyer, D.M., Wood, A.R., Kraus, M.J., McInerney, F.A., Krigbaum, J., 2012. Evolution of the earliest horses driven by climate change in the Paleocene-Eocene Thermal Maximum. Science 335, 959e962. Seiffert, E.R., Boyer, D.M., Costeur, L., 2015. Tarsal morphology of Caenopithecus, a large adapiform primate from the middle Eocene of Switzerland. PeerJ 3, e1036. Seiffert, E.R., Perry, J.M.G., Simons, E.L., Boyer, D.M., 2009. Convergent evolution of anthropoid-like adaptations in Eocene adapiform primates. Nature 461, 1118e1121. Seiffert, E.R., Simons, E.L., Attia, Y., 2003. Fossil evidence for an ancient divergence of lorises and galagos. Nature 422, 421e424. Soligo, C., Martin, R.D., 2006. Adaptive origins of primates revisited. Journal of Human Evolution 50, 414e430. Soligo, C., Müller, A.E., 1999. Nails and claws in primate evolution. Journal of Human Evolution 36, 97e114. Styner, M., Oguz, I., Xu, S., Brechbuler, C., Pantazis, D., Gerig, G., 2006. Framework for the statistical shape analysis of brain structures using SPHARM-PDM. Insight Journal 1071, 242e250. Szalay, F.S., 1976. Systematics of the Omomyidae (Tarsiiformes, Primates) taxonomy, phylogeny, and adaptations. Bulletin of the American Museum of Natural History 156, 157e450. Vitek, N.S., Manz, C.L., Gao, T., Bloch, J.I., Strait, S.G., Boyer, D.M., 2017. Semi-supervised determination of pseudocryptic morphotypes using observer-free characterizations of anatomical alignment and shape. Ecology and Evolution 7, 5041e5055. von Koenigswald, W., Habersetzer, J., Gingerich, P.D., 2012. Pedal distal phalanges of the Eocene adapoids Europolemur and Darwinius compared to phalanges of Notharctus and other primates. Palaeobiodiversity and Palaeoenvironments 92, 539e565. Williams, B.A., Covert, H.H., 1994. New early Eocene anaptomorphine primate (Omomyidae) from the Washakie Basin, Wyoming, with comments on the phylogeny and paleobiology of anaptomorphines. American Journal of Physical Anthropology 93, 323e340. Williams, G.C., 1966. Adaptation and Natural Selection. Princeton University Press, Princeton. Wing, S.L., Harrington, G.J., Smith, F.A., Bloch, J.I., Boyer, D.M., Freeman, K.H., 2005. Transient floral change and rapid global warming at the Paleocene-Eocene boundary. Science 310, 993e996.

Please cite this article in press as: Boyer, D.M., et al., Oldest evidence for grooming claws in euprimates, Journal of Human Evolution (2018), https://doi.org/10.1016/j.jhevol.2018.03.010