Patterns in alligatorine evolution

Transcription

1 University of Iowa Iowa Research Online Theses and Dissertations Fall 2016 Patterns in alligatorine evolution Jessica Miller-Camp University of Iowa Copyright 2016 Jessica Miller-Camp This dissertation is available at Iowa Research Online: Recommended Citation Miller-Camp, Jessica. "Patterns in alligatorine evolution." PhD (Doctor of Philosophy) thesis, University of Iowa, Follow this and additional works at: Part of the Geology Commons

2 PATTERNS IN ALLIGATORINE EVOLUTION by Jessica Miller-Camp A thesis submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree in Geocience in the Graduate College of The University of Iowa December 2016 Thesis Supervisor: Professor Christopher A. Brochu

3 Copyright by JESSICA MILLER-CAMP 2016 All Rights Reserved

4 Graduate College The University of Iowa Iowa City, Iowa CERTIFICATE OF APPROVAL PH.D. THESIS This is to certify that the Ph.D. thesis of Jessica Miller-Camp has been approved by the Examining Committee for the thesis requirement for the Doctor of Philosophy degree in Geoscience at the December 2016 graduation. Thesis Committee: Christopher A. Brochu, Thesis Supervisor Ann Budd Bradley Cramer John Logsdon Alan Turner

5 ACKNOWLEDGMENTS This research was funded by multiple sources. Internally, the UI Department of Earth and Environmental Sciences provided travel grants to conferences and monies from the Littlefield Fund for research. ECGPS (now GPSG) provided a student research grant. The T. Anne Cleary International Dissertation Research Fellowship made many visits to international collections possible. Externally, The Evolving Earth Foundation provided a student research grant. The Paleontological Society Kenneth E. and Annie Caster Award also funded this research. Additional time focused on research was made possible by a Graduate College Summer Fellowship and SIF Fellowship Many museum personnel provided access to their collections. They are: Carl Mehling at AMNH; Sue McLaren at CM; Blaine Schubert, Stephen Wallace, and Brett Woodward at ETMNH; Richard Hulbert, Kenneth Krysko, and Alex Hastings at FMLNH; Bill Simpson and Jonathan Mitchell at FMNH; Meinholf Hellmund and Alex Hastings at GMH; Thomas Schossleitner and Daniela Schwarz-Wings at HMN; Annelise Folie at IRSNB; Jun Liu at IVPP; Martin New and Alex Peaker at IWCMS; Linda Ford, Michelle Kennedy, and Farish Jenkins at MCZ; Löic Costeur at MH; Ronan Allain at MHN; Stéphane Jouve at MHNM, who also provided contact with and transport to MHNA and a private collection nearby, and thanks to both him and his wife, Marie, for graciously providing me lodging and food; Jin Jianhua and his two students at MMC; Vince Schneider at NCSM; Lorna Steele at NHMUK; David Parris at NJSM; Justin Spielmann at NMMNH; Michael Brett Surman at NMNH; Andy Farke at RAM; Komsorn Lauprasert and many staff members and students at SM and Maha Sarakham University; Krister Smith at SMF; Kenny Bader, Chris ii

6 Sagebiel, Matt Brown, and Tim Rowe at TMM; Pat Holroyd at UCMP; Ron Eng, Regan Dunn, and Chris Sidor at UWBM; Daniel Brinkman and Christopher Norris at YPM. Photographs of additional specimens were provided by Stephanie Drumheller Horton, Rudd Sadler, Chris Brochu, and Yanyin Wang. In addition to people mentioned above, discussion was provided by other members of the Brochu lab group, Brad Cramer, and Pete Sadler. Much moral support was provided by Bob Miller Camp, Chris Brochu, Andy Heckert, and many others. iii

7 ABSTRACT Alligatorines are a diverse clade of crocodylians whose history spans the entire Cenozoic. They are suited to answer a variety of questions with far reaching impacts due to their physiology and preservation potential, and have been the subject of several phylogenetic, biogeographic, and diversity analyses. However, prior phylogenetic analyses had poor resolution and several putative alligatorines have never been included, while other analyses would be more informative and accurate if viewed through the context of evolutionary history. Here, I analyze the phylogenetics, taxonomy, biogeography, ecomorphology, and diversity dynamics of alligatorines. An almost fully resolved phylogenetic hypothesis returns two major clades within Alligatorinae and includes several putative alligatorines not previously analyzed. The clade originated in North America and dispersed to Europe and Asia three to five times via at least three different corridors at high latitudes when climate and potentially salinity were favorable, likely including the recently discovered subaerial Lomonosov Ridge. The modern American alligator is a dietary generalist, but evolved from a durophagous specialist, contrary to the intuitive reasoning of the Law of the Unspecialized. It was able to do so by entering the generalist niche vacated by basal crocodyloids following their extirpation from mid latitude North America. Alligatorine diversity only weakly tracks climate change and does not track the rock record excepting swampy environments. Alligatorine diversity correlates with climate change. Climate change correlates with rocks, though in a more complicated pattern. Some diversity metrics correlate with some aspects of the rock record, but predominantly do not. There is more support for the common cause hypothesis than for rock record bias driving apparent alligatorine diversity. Overall, alligator evolution exhibits a pattern of being more diverse taxonomically and morphologically when the climate is warmer, and dispersing during the warmest and wettest of those times. iv

8 PUBLIC ABSTRACT Alligatorines the American alligator and every species more closely related to it than to the spectacled caiman are a diverse clade of crocodylians with a fossil record going back almost 66 million years, the time of the K Pg mass extinction. They are suited to answer a variety of questions with far reaching impacts due to physiological features such as salt intolerance and cold bloodedness, as well as their high preservation potential in the rock record. They have been the subject of several analyses studying how they re related to one another, how they came to be in the geographic regions they re found in, and how their diversity has changed over time. However, prior analyses of how the various species are related included many inconclusive results and several potential alligatorines have never been included. And analyses of biogeography and diversity would be more informative and accurate if viewed through the context of evolutionary history. Here, I analyze their family tree, biogeography, how their shape changes based on what they re eating, and how their diversity correlates (or doesn t) with climate change. An almost fully resolved family tree shows two groups within Alligatorinae and includes several species not previously analyzed. Alligatorinae originated in North America and dispersed to Europe and Asia between three to five times via at least three different paths at high latitudes when climate and potentially ocean salinity were favorable, likely including the Lomonosov Ridge, which today crosses the North Pole and is entirely underwater. The modern American alligator is a dietary generalist, eating anything and everything it can. But it evolved from species specialized to crack hard shells. This evolutionary direction runs contrary to the intuitive reasoning of the Law of the Unspecialized, which states that specialists go extinct when the environment shifts, while generalists are able to make it through unfavorable changes by being more adaptable. Alligators were able to break it by entering the generalist dietary niche vacated by now extinct crocodile relatives after the latter went extinct in mid latitude North America due to their inability to tolerate cold. Rock record bias, wherein apparent diversity is v

9 due to rock exposure rather than actual diversity signals, should always be considered as a possible influence in diversity studies. This is fortunately not the case for alligators, with diversity tracking climate rather than rocks. Alligators are more diverse in number of species and in range of ecological niches, as well as being more likely to disperse, in warmer, wetter times than in cooler, drier times. vi

10 TABLE OF CONTENTS LIST OF FIGURES... ix LIST OF TABLES... xi CHAPTER I. PHYLOGENY AND TAXONOMY...1 Introduction...1 Taxonomic Review...3 Allognathosuchus and Species Sometimes Referred to it...3 Alligator...8 Asian Alligatorids...9 Materials and Methods...11 Results...13 Discussion...14 II. BIOGEOGRAPHY...17 Introduction...17 Materials and Methods...19 Results...21 Discussion...23 Lagrange...23 Climate...24 Dry Land Corridors...25 Salinity...27 III. ECOMORPHOLOGY...30 Introduction...30 Materials and Methods...31 Results...33 Discussion...35 IV. DIVERSITY AND CLIMATE CHANGE...40 Introduction...40 Materials and Methods...41 Results...43 Diversity...43 Rock Record...43 Climate...45 Discussion...46 V. SYNTHESIS...48 Introduction...48 Chapter 1: Phylogenetics and Taxonomy...48 vii

11 Chapter 2: Biogeography...49 Chapter 3: Ecomorphology...50 Chapter 4: Diversity Dynamics...52 Conclusions...53 Future Directions...53 APPENDIX A. INSTITUTIONAL ABBREVIATIONS...55 APPENDIX B. CHARACTER LIST AND CHARACTER STATE DESCRIPTIONS...57 APPENDIX C. CHANGES TO CHRARACTER LIST FROM PREVIOUS MATRICES...72 APPENDIX D. TAXON-CHARACTER MATRIX FOR THE PHYLOGENETIC ANALYSIS AND MPTS...78 APPENDIX E. LAGRANGE FILE OF SPECIES IN BIOGEOGRAPHIC AND DIVERSITY ANALYSES...98 APPENDIX F. SPECIMENS AND MEASUREMENTS IN MORPHOMETRIC ANALYSES APPENDIX G. COUNTS OF ALLIGATORINE DIVERSITY AND ROCK EXPOSURE REFERENCES viii

12 LIST OF FIGURES Figure 1. Alligator sinensis (left) and Alligator mississippiensis (right), the extant alligatorines. Photo credit to Stephanie Drumheller-Horton Globidontan alligatoroids from Brochu, A, UCMP , Brachychampsa montana; B, RTMP , holotype, Stangerochampsa mccabei; C, AMNH 6780, holotype, Allognathosuchus mooki; D, MCZ 8381, cf. Allognathosuchus mooki; E, AMNH 5186, holotype, Navajosuchus novomexicanus; F, HLMD 4415, holotype, Allognathosuchus haupti; G, YPM-PU 16989, Willwood alligatorid (identified as Allognathosuchus wartheni in Brochu, 1999); H, AMNH 6049, Allognathosuchus polyodon; I, SMM p , holotype, Wannaganosuchus brachymanus; J, MNHN QU17155, holotype, Arambourgia gaudryi; K, CM 9600, holotype, Procaimanoidea kayi; L, USNM 15996, holotype, Procaimanoidea utahensis; M, SDSM 243, Alligator prenasalis; N, Alligator mcgrewi, composite reconstruction based on the holotype (FMNH p26242) and AMNH 7905; O, TMM m-7487, Alligator mississippiensis Top from left to right: A. sinensis, A. cf. sinensis, Krabisuchus, A. luicus. Bottom: Maoming gator Strict consensus trimmed to Alligatorinae. Six full MPTS in Appendix D The California alligatorine, sp. nov Map from Briakatis 2014 showing the location of the DeGeer (1), Thulean (2), and Beringian (3) land bridges Map from Brinkuis 2006 showing the location of the Lomonosov Ridge and documented Azolla blooms (stars) Globidont teeth in various taxa. Top row is: mosasaur, savannah monitor, placedont. D-F are globidont teeth in fossil alligatorines. E,F show loss in modern American alligator Snout length and size proxy measurements Growth trajectories in eusuchians and derived neosuchians Ancestral state reconstruction of snout length relative to size. Colors range continuously and indicate brevirostry (dark blue) to longirostry (red) Largest size sampled in each species as demonstrated by the anterior skull table width proxy. Red=gigantic, dark blue=very small Co-occurrence of alligatorines and other crocodylian and choristodere species and their snout lengths ix

13 14. Cenozoic climate change. Image source: Alligatorine biodiversity through time. Orange=raw counts; Green=inclusion of ghost lineages and range extensions; purple=ghost lineages, range extension, and unincluded taxa tacked on Rock exposure through time. Orange=swamp; Green=lacustrine, Purple=fluvial, Black=total δ 18 O values for most of the Cenozoic. Note that values of the proxy are inversely correlated with temperature, so the y-axis has been flipped such that hugher temperatures are up x

14 LIST OF TABLES Table 1. Spearman's rho correlation coefficients, color-coded to range from dark green (strong negative correlation) to dark orange (strong positive correlation) xi

15 1 CHAPTER I: PHYLOGENY AND TAXONOMY Introduction Alligatorinae is the clade consisting of the American alligator (Alligator mississippiensis) and all crocodylians more closely related to it than to the spectacled caiman (Caiman crocodilus) (Brochu 1999b). Only two species are extant, A. mississippiensis and the Chinese alligator (A. sinensis), but the clade was much more diverse in the past. Alligatorines arose shortly before or after the Cretaceous Paleogene (K Pg) boundary and were widespread in mid to high latitudes of North America, Europe, and Asia (Densmore and Owen 1989; Hass et al. 1992; Brochu 1999a; Janke et al. 2005; Brochu 2011a, b; Oaks 2011). They survived largescale climate changes, dispersed between continents multiple times, and their ecomorphology drastically changed during the course of their evolution. Like all crocodyliformes, crocodylians and their extinct relatives, alligatorines have been viewed as living fossils primitive relicts of the past that have undergone little change throughout their history (Buckland 1836). This is a misperception, as the alligatorine record reveals a remarkably diverse and phylogenetically complex group. The earliest forms were small (<2 m) and had anvil like (globidontan) cheek teeth (Mook 1921; Case 1925; Patterson 1931; Bartels 1988; Brochu 1999a, 2004a). Others had deeper snouts and may have spent less time in the water (Brochu 2004a). The most widespread living alligtorine, A. mississippiensis, is a model species in many genomic, physiological, ecological, taphonomic, and comparative morphological research (see Grigg and Krishner, 2015). Figure 1: Alligator sinensis (left) and Alligator mississippiensis (right), the extant alligatorines. Photo credit to Stephanie Drumheller- Horton. Crocodyliforms have been an increasingly popular topic of paleontological research in recent years. Many questions remain regarding their evolution, due in large part to the poorly resolved and supported phylogenetic hypotheses that have been put forward

16 2 thus far (Wu et al. 1996; Brochu 1999b, 2004b; Snyder 2007; Martin and Lauprasert 2010b; Brochu 2011a; Brochu 2011c; Whiting 2014). Did A. mississippiensis a large ecological generalist evolve from smaller, more specialized durophagous ancestors? How diverse were the radiations of basal alligatorines and early species of Alligator? How and when did alligatorines acquire their current disjunct distribution? What factors have driven alligatorine diversity over time, and how strongly did each factor contribute to that pattern? How exactly are the various species related to one another? Complete genomes for both living alligatorines are available (St. John et al. 2012; Wan et al. 2013). This, along with the rich fossil record for the group, presents a unique opportunity to compare molecular and paleontological approaches to understanding evolutionary rate, divergence time, and phylogenetic responses to well calibrated climatic and tectonic changes throughout the Northern Hemisphere. But phylogenetic analyses including all known species have not been conducted, and many species remain undescribed due largely to the poor preservation of some species, lack of attention by crocodylian workers, and the recent discovery of completely new species (Brochu 1999, 2004, 2011a, b). Alligatorine phylogeny has been examined in several studies (Brochu 1999, 2004, Martin 2010, Wu 1996). Many of these studies use the Brochu matrix either alone or as the basis for their own. His matrix was first published in 1997 and consisted of 164 characters. It was updated in subsequent publications, adding 15 characters and modifying some states (Brochu 2011a, 2011b). However, my observations of specimens in collections have indicated that the matrix is in need of a major overhaul in order to more accurately account for the degree of intraspecies variation present, to delineate differing states which were unaccounted for in the past, and to include characters not previously considered. A major collaboration between the labs of Chris Brochu (Univ. of Iowa) and Alan Turner (Univ. of Stony Brook) is underway to combine and update published crocodylomorph matrices. A preliminary version of that is used in this study, with both the lab leaders and myself contributing new characters and states, as well as rewording

17 3 states in ways that make them more straightforward for other researchers. New characters have been added coding for observed variation in the cranium and postcrania, the latter having received little attention in past analyses of Alligatorinae due in part to the assumption that crocodylian postcranial anatomy does not differ substantially in the crown group, and in part because crania tend to be more well preserved. This dissertation will explore alligatorine evolution and the larger questions they are suited to answer. But first an overview of parts of alligatorine taxonomy is needed as so many new species have come to light since the last overviews a decade and a half ago (Brochu 1999, 2004). Taxonomic Review Allognathosuchus and Species Sometimes Referred to it Historically, most Paleogene alligatorines the majority of which are small and probably durophagous were referred to Allognathosuchus (Mook 1921; Abel 1928; Patterson 1931; Bartels 1988; Brochu 1999a, 2004a; Lucas, 1992; Lucas and Estep, 2000). The genus was reported from California to Germany and was thought to be present on both sides of the K Pg boundary. Later work showed that Allognathosuchus is a wastebasket taxon a taxon which many species are assigned to based on a priori assumptions or due to initial ignorance of distinguishing characteristics and some named species are based on insufficient material (Brochu 2004a), but not all of the formations with specimens called Allognathosuchus sp. have been evaluated. Some of these specimens may represent new species or range extensions (either temporal or geographic) of other species previously assigned to Allognathosuchus. Clarification of this assemblage is needed both because of the basal position many putative Allognathosuchus have adopted in phylogenetic analyses and because of the widespread occurrence of Allognathosuchus like alligatorines throughout the Northern Hemisphere during the Paleogene (Brochu 2004a). I was able to view and study some potential new species. I have included one, from the middle Eocene Santiago Formation of California, in my phylogenetic analysis here. A description of this species has been written and will not be

18 4 Figure 2: Globidontan alligatoroids from Brochu, A, UCMP , Brachychampsa montana; B, RTMP , holotype, Stangerochampsa mccabei; C, AMNH 6780, holotype, Allognathosuchus mooki; D, MCZ 8381, cf. Allognathosuchus mooki; E, AMNH 5186, holotype, Navajosuchus novomexicanus; F, HLMD 4415, holotype, Allognathosuchus haupti; G, YPM-PU 16989, Willwood alligatorid (identified as Allognathosuchus wartheni in Brochu, 1999); H, AMNH 6049, Allognathosuchus polyodon; I, SMM p , holotype, Wannaganosuchus brachymanus; J, MNHN QU17155, holotype, Arambourgia gaudryi; K, CM 9600, holotype, Procaimanoidea kayi; L, USNM 15996, holotype, Procaimanoidea utahensis; M, SDSM 243, Alligator prenasalis; N, Alligator mcgrewi, composite reconstruction based on the holotype (FMNH p26242) and AMNH 7905; O, TMM m-7487, Alligator mississippiensis. included here (Hutchinson, in prep.). I also included the Willwood Alloganthosuchus, which has, in turns, been considered valid, a junior synonym of another Allognathosuchus species, or to actually be two species. A cast of a jaw from Ellesmere Island is unfortunately too fragmentary to be included in this analysis. Lastly, I consider a third,

19 5 Allognathosuchus heterodon, which may be a junior synonym of Alloganthosuchus polyodon (Brochu 1999, 2004). Two species of Allognathosuchus are considered valid by all authors, A. polyodon and the Willwood alligatorid. Allognathosuchus polyodon is the type species (Cope 1873). Allognathosuchus wartheni (Case 1925) has been called the Willwood alligatorid ((Lucas 1992; Lucas and Estep 2000; Brochu 2004b; Brochu 2011a; Brochu 2011c)). The presence of a species unique from A. polyodon has not been disputed, but its name and exact identity has, with some authors asserting the presence of two species in the Willwood Fm (Bartels 1980; Bartels 1988). The holotype is unfortunately fragmentary; what is preserved is not distinguishable from other species of Allognathosuchus according to several authors (Case 1925; Lucas 1992; Lucas and Estep 2000), and would thus be a nomen dubium. I also observed two different posterior tooth row morphologies from the Willwood Formation assigned to A. wartheni. Brochu (1999, 2004) considered Allognathosuchus heterodon to be synonymous with the Willwood alligatorid. Some specimens from the Willwood Fm. are distinguishable from A. heterodon, and others are not. Some specimens previously considered to be A. wartheni should likely be subsumed into A. heterodon. I have retained a distinction between A. wartheni and A heterodon to consider the distinct morph, with only specimens distinguishable from A. heterodon used to code the distinct Willwood alligator. For the purposes of this dissertation, I have continued to call this taxon A. wartheni or the Willwood alligator, but note that a name change may be forthcoming. Several other species previously assigned to Allognathosuchus have been returned in more basal positions in the quantitative phylogenetic analyses that have included it (Brochu 1999b, 2004b; Brochu 2011a; Brochu 2011c). They had also been called by different generic names (Navajosuchus and Hassiacosuchus) by previous authors, so those names have since been resurrected (Brochu 1999b). Navajosuchus itself has had two species named, N. mooki (the type, originally called Allognathosuchus mooki) and N. novomexicanus, but the latter has been considered a junior synonym by all recent authors (Sullivan et al. 1988; Lucas 1992; Brochu 1999b, 2004b; Brochu 2011a; Brochu 2011c). The exclusion of Navajosuchus mooki from

20 6 Allognathosuchus has been challenged by a few authors (Lucas and Estep 2000; Lucas and Sullivan 2004). However there are numerous differences between Navajosuchus and the Willwood alligatorid which Lucas, Estep, and Sullivan (Sullivan et al. 1988; Lucas and Estep 2000; Lucas and Sullivan 2004) believe it to be synonymous, and these are the only recent workers who assert it as such. Among these differences are the presence of a smaller incisive foramen in at least some Willwood alligatorid specimens than in Navajosuchus. PU 16989, for instance, possesses an incisive foramen roughly 13% the width across the premaxillae, while the measurement Lucas and Estep (2000) made for the N. mooki holotype is 30%. The tooth count is also different between the two species. If an argument is made that like skull sizes should be compared, assuming these two to be the same species would require the acquisition of more teeth during ontogeny. I have observed tooth loss during ontogeny due to the rapidly increasing size of some alveoli (e.g., in Crocodylus niloticus), but not acquisition. Hassiacosuchus haupti is from the middle Eocene Messel Pit in Germany. Berg (1966) considered it a species of Allognathosuchus. The consensus among other authors based on descriptive and phylogenetic work is that it is not Allognathosuchus. There have been two other blunt snouted species named from Europe Caimanosuchus brevirostris and Eocenosuchus weigelti, which Brochu (1999) considered to possibly be in the same genus, though he left the possibility for their validity open. I have not been able to observe a difference between them that cannot be attributed to ontogenetic variation, and thus I consider them synonymous with Hassiacosuchus haupti. Another species which has been treated variously as a species of Allognathosuchus (Sullivan et al. 1988; Lucas and Estep 2000) and as a distinct genus (Brochu 1999, 2004,2011a,2011b; Wu 1996; Erickson 1982) is Wannaganosuchus brachymanus. Lucas and Estep (2000) assert that Wannaganosuchus is a junior synonym based on the following attributes, which are the only ones Wu et al. (1998) identified as unique: posterior extent of the nasals, relative lengths of the nasal premaxilla and nasal maxilla sutures, and relative lengths of the anterior and posterior processes of the jugal. Wu et al. (1998) perceived the nasals to end anterior to the orbits in Allognathosuchus but at or posterior

21 7 to them in Wannaganosuchus. Lucas and Estep (2000) asserted that they end posterior to the anterior ends in both genera. Wu et al. (1998) stated that the sutures the nasal has with the premaxilla and maxilla are the same length in both, whereas Lucas and Estep claim the nasal maxilla contact to be longer in both. Lastly, Wu et al. (1998) claim the two jugal processes differ in length in Wannaganosuchus, while Lucas and Estep (2000) claim they are the same length in both. My observations on these traits concur with those of Lucas and Estep (2000). Brochu (1997, 1999) identified three additional features unique between Wannaganosuchus and Navajosuchus. The first is the dorsally directed opening of the naris in Wannaganosuchus, and anterodorsal in Navajosuchus. Lucas and Estep (2000) observed that the naris opens anterodorsally in both species. My observations corroborate this for coding purposes, though the angle between the two is noticeably different. The second is that the lacrimal is longer than the prefrontal in Navajosuchus and vice versa for Wannaganosuchus (Brochu 1997; Brochu 1999b) or that the prefrontal is longer in both (Lucas and Estep 2000). Again, my observations are in agreement with those of Lucas and Estep (2000), though the difference in length is slight and would perhaps be better coded in a matrix as the same or nearly the same length. The last trait, the width of the incisive foramen relative to the greatest width across the premaxillae, is stated to be more than half in Navajosuchus, but less than half in Wannaganosuchus by Brochu (1997; 1999) and less than half in both species by Lucas and Estep (2000). My observations are in agreement with that of Lucas and Estep (2000). I am not in agreement with Lucas and Estep (2000) that the two are the same species. Two characters have been identified subsequent to their paper that distinguish the two species. Wannaganosuchus has a maxillary ramus of the ectopterygoid that makes up less than two thirds of the lateral border of the suborbital fenestra, whereas it makes up more than two thirds in Navajosuchus (Brochu 2004b). The surangular has substantial participation on the dorsal surface of the retroarticular process in Navajosuchus, but minimal participation in Wannaganosuchus. Furthermore, it is readily apparent that Wannaganosuchus has a relatively short snout and wide skull, while specimens of

22 8 Navajosuchus have relatively longer and more acute snouts regardless of size. I also disagree with Lucas and Estep (2000) on the ontogenetic stage of Wannaganosuchus, which they believe to be a juvenile Allognathosuchus. The skull roof of the Wannaganosuchus type, and only specimen confirmed to belong to it, is wider posteriorly than anteriorly. In all my observations of numerous crocodylian skulls including complete ontogenetic series of modern specimens whose life stages were known there have been no juvenile individuals for whom that was true. Crocodylians begin life with a skull roof whose sides are either parallel or wider anteriorly. As they age, the posterior width increases faster than the anterior width, resulting in the adult state of having a wider posterior width just as is present in the type and only specimen of Wannaganosuchus. Though there are currently few differences between the two species in matrices, I believe that to be due to the lack of sufficient characters in matrices distinguishing them rather than a true absence of distinction. Several other taxa have been referred to Allognathosuchus. Rauhe (1990) considered Arambourgia gaudryi, Caimanosuchus brevirostris, and Eocenosuchus weigelti to all be junior synonyms of Hassiacosuchus haupti, which was then considered to be Allognathosuchus. I have not observed compelling evidence to the contrary for the latter two and agree that they are synonymous with Hassiacosuchus, but Arambourgia is substantially different both in codings and in general appearance, so I follow Brochu (1999, 2004) in treating them as separate species. Lucas (1992) considered Albertochampsa langstoni to be a junior synonym based on a single feature he says is variable in Allognathosuchus; however, given his assertions that species consistently recovered outside Alligatorinae, much less outside of Allognathosuchus, in quantitative phylogenetic analyses are still part of it, this further assertion is not supported. Alligator A 20 million year gap separates the earliest species of Alligator (A. prenasalis) in the late Eocene from the closest relative of Alligator, late Paleocene Wannaganosuchus (Brochu 2004a, 2011a, b). A morphological gap also separates these forms;

23 9 Wannaganosuchus is small and bears bulbous cheek teeth like those of earlier alligatorines, but A. prenasalis lacks crushing dentition and has a longer snout. A new species from the Uintan fills this gap (Brochu et al. in prep). I have included it in my analyses but will not be describing it here. At present, five Neogene species of Alligator are accepted, one of which (A. luicus from China) has not yet been included in phylogenetic work. Most are from the western plains (Nebraska, South Dakota) and the southeast of the United States. Additional material is known from Tennessee and Florida that may represent new species (Schubert et al., in prep; not included in this analysis). Asian Alligatorids The putative Asian alligatorids warrant a separate discussion due to the recent findings of potential new species and the lack of attention previously known specimens have received. The species and specimens up for discussion are: Alligator sinensis and fossils assigned to it, Alligator luicus, Eoalligator, Krabisuchus, the Maoming gator, and Lianghusuchus. Figure 3: Top from left to right: A. sinensis, A. cf. sinensis, Krabisuchus, A. luicus. Bottom: Maoming gator Alligator sinensis is the extant Chinese alligator. Due to prior familiarity, fossils as old as the Miocene have been assigned to it. Unfortunately, most of these fossils are fragmentary isolated teeth or osteoderms. Because they are nondiagnostic, the possibility that the older ones represent a now extinct species cannot be ignored. At best, these scraps can be identified as Alligatoridae sp. A recent unpublished find in Japan, while less fragmentary, also falls in this category at least if the images published in an outreach handout are representative of the quality of preservation.

24 10 Two other recent finds, however, are of nearly complete skulls with associated postcrania. The first of these is from the Penghu Channel between the PRC and Taiwan (Hsi yin 2013). It is Pleistocene in age and its features are consistent with the modern species. I concur with its current taxonomic assignment as such. The second one looks very different and is substantially older. It is from earlier in the Pleistocene of Thailand (Claude 2011). It is currently coated in rock, so the sutures are not visible, but its overall shape is that of a much deeper skull with a much shorter snout than that possessed by A. sinensis. Its current taxonomic assignment is A. cf. sinensis, and, while it s possible that it could represent an early member of the species, it is improbable based on the amount of intraspecific variation observed in modern species. It is much more likely to represent a new species related to the modern taxon. One alligatorine is known from the Miocene of Asia Alligator luicus. It is represented by a single juvenile individual that is unfortunately flattened and embedded in translucent epoxy. Its preservation is such that its features are difficult to distinguish. Two alligatorines are known from the Eocene of Asia Krabisuchus and the Maoming alligator. Krabisuchus is from southern Thailand. While material uncovered thus far has been partial and crushed or distorted, more continues to be excavated, so it will hopefully become better known over time. Its exact relationship to other alligatoroids is uncertain as it has been uncovered as a more basal alligatoroid in one analysis and has always been recovered in polytomies. This seems to be due to an unusual number of autapomorphies. It does possess a laterally shifted quadrate foramen aerum, a trait unique to alligatoroids, so there is little doubt that it does belong somewhere in this clade. The Maoming alligator is from Guangdong Province in South China. It is unfortunately poorly preserved, with a flattened skull that is only exposed dorsally, and with some of the surface bone prepped away, and a previously unpublished cervical and partial thoracic vertebral series. Lianghusuchus is known from some maxillary, quadratojugal, vertebral, and osteoderm fragments (Young 1948). Prior to Brochu (1999), any mention of it was as a crocodyloid, and it was treated as such in a compendium of fossil reptiles from China (Li

25 11 et al. 2008). However, Brochu (1999) noted that the morphology of its maxillary tooth row was more suggestive of an alligatoroid. It has been reported as being Eocene in age and is geographically very close to the Maoming alligator (Young 1948). This would appear to suggest that the two are one and the same, but a magnetostratigraphic analysis has indicated that the lower fossiliferous layer of the Lingcha Fm, where Lianghusuchus is from, is actually latest Paleocene in age, as the Paleocene Eocene isotopic excursion is present in the upper fossiliferous layer. If Lianghusuchus truly is an alligatoroid, as its maxillary tooth row morphology suggests, then it would be the oldest record of an alligatoroid in the Old World. I have not seen the material and the published drawings do not lend themselves to coding, so I have not included it in the phylogenetic analysis, but discuss how it could change results later if it is actually an alligatorine. The remaining putative alligatorids, Eoalligator chunyii and Eoalligator huiningensis, do not appear to actually be alligatorids. Their morphology is much more indicative of their being basal crocodyloids (particularly the large notch between premaxilla and maxilla). This is even more apparent from the photograph of a new complete skull from Anhui which is currently being worked on (Liu and Wang, pers. comm.) and is probably E. huiningensis given its age and location. E. chunyii has recently been synonymized with the basal crocodyloid Asiatosuchus nanlingensis (Wang et al., 2016). This study used the previous version of Brochu s matrix and recovered E. huiningensis as an alligatoroid, renaming it Protoalligator huiningensis. However, it is possible the three known skulls of Eoalligator represent a single species. E. chunyii is fragmentary, but consistent with the other two in morphology, time, and relatively nearby location. I have coded E./P. huiningensis in this study to determine if its recovered topology changes or stays the same using the updated matrix employed here. Materials and Methods I examined eusuchian specimens in 29 collections (Appendix A) over the course of this project in addition to photographs of additional specimens I did not see in person. I included 114 species in the phylogenetic analysis with Bernissartia fagesii as the outgroup.

26 12 I did not have images or personally see all of the non alligatoroid species, so some of them have not been coded for the new characters. The list of 218 characters (Appendix B) is a current incarnation of a matrix being compiled by the Brochu and Turner labs. In its current state, it is based mostly on the Brochu (2011b) matrix, with some changes/additions from the Turner (2015) matrix. There are many new characters and states, and some old characters and states have been removed (Appendix C). Fifteen characters (2, 8, 14, 16, 26, 27, 43, 45, 47, 98, 100, 105, 109, 120, and 176) are not included in the phylogenetic analysis in this study. The final states of these characters have not been determined yet, but they have been retained in the character list to minimize potential changes in character numbers in later incarnations of this matrix. I ran a maximum parsimony analysis in TNT v. 1.1 (Goloboff 2008). All characters were equally weighted. One character (185) was ordered. This character is: Parietal and squamosal widely separated by quadrate on posterior wall of supratemporal fenestra (0); parietal and squamosal approach each other on posterior wall of supratemporal fenestra without actually making contact (1); parietal and squamosal meet along posterior wall of supratemporal fenestra (2). The second state is clearly intermediate. While many authors are staunchly against ordered characters as a matter of minimizing a priori assumptions, for characters with obviously progressive states, failing to acknowledge the obvious intermediacy of one or more states is akin to selectively ignoring data. If characters such as this are not ordered, the analysis will treat all shifts equally, when in actuality the shift from one end member to the other is a larger change than one step shifts along the continuum. I found the most parsimonious trees (MPTs) using a new technology search with ratchet, drift, and tree fusing on default settings and 200 random addition sequences. To assess node support, I calculated nonparametric bootstrap proportions (1000 replicates, new technology search with ratchet, drift, and tree fusing on default settings) using GC frequencies and decay indices (determined on all suboptimal trees five steps longer than the MPTs) in TNT (Felsenstein 1985).

27 13 Results The phylogenetic analysis resulted in six MPTs with a tree length (TL) of 937, and TL of Alliatorinae only was 115 (Appendix D). I used a strict consensus to summarize the results (Figure 4). The resulting phylogeny as returned by the strict consensus tree is well resolved, but support is low within Alligatorinae. Only two Alligator clades are well supported. The topology of the strict consensus is almost completely congruent with published analyses of Alligatorinae. The Maoming alligator, Chrysochampsa, and Alligator luicus the three alreadypublished alligatoroid taxa not previously included in a published phylogenetic analyses (in the case of Chrysochampsa, due to wild card behavior [Brochu 1999] and inclusion of the other two simply had not been attempted yet) had stable positions, albeit surprising in the case of the Maoming alligator. Eoalligator had to be removed as it caused many clades to collapse, but was recovered as a basal crocodyloid in the preliminary phylogenetic analysis, as expected. Figure 4: Strict consensus trimmed to Alligatorinae. Six full MPTS in Appendix D.

28 14 Discussion The polytomy present at the base of Alligatorinae in published analyses is fully resolved, with Navajosuchus and Ceratosuchus in a sister taxon relationship separate from other alligatorines. Hassiacosuchus has been shifted up one node to be the basalmost taxon inside a clade I am calling Allognathosuchini. Even though there is low support, Allognathosuchus, Arambourgi, and Procaimanoidea consistently form a clade distinct from Alligator + Wannaganosuchus in analyses. I am defining it as Allognathosuchus polyodon and all taxa closer to it than to Alligator mississippiensis. Within Allognathosuchini, there are two subclades: one consisting of Krabisuchus and Allognathosuchus, the other containing Procaimanoidea, Arambourgia, and the CA Figure 5: The California alligatorine, sp. nov. Allognathosuchus which is not actually Allognathosuchus as it does not form a monophyletic clade with them. A. wartheni is returned as the sister axon of A. polyodon and has no autapomorphies. The CA Allognathosuchus is the basalmost taxon in its subclade and Procaimanoidea is monophyletic. I am calling the sister taxon of Allognathosuchini in this analysis Alligatorini and defining it as Alligator mississippiensis and all taxa closer to it than to Allognathosuchus

29 15 polyodon. The basalmost taxon is Wannaganosuchus, followed by the Uintan gator that is currently being described (Brochu et al., in prep), then a monophyletic Alligator. The basalmost Alligator species, A. prenasalis and A. mcgrewi, form a polytomy with a clade containing the remaining species. This is the only section of the alligatorine tree which is less resolved than previous analyses. In prior publications, A. mcgrewi was crownward of A. prenasalis. This is more in keeping with the stratigraphic record, but that relationship has been lost here. A single unambiguous character had united all Alligator except A. prenasalis loss of supratemporal exposure on the skull roof table. The inclusion of Alligator luicus has resulted in multiple most parsimonious topologies in this part of the tree, as it appears to retain dorsal exposure of its supraoccipital. It is not impossible the valley I have interpreted as the supraoccipital parietal suture is a confluence of rugosity many of the specimen s sutures are hard to distinguish but given its similarity to its frontoparietal suture, I have coded it as a suture here. Alligator luicus is the next species to branch off, followed by a well supported clade of crown Alligator + A. olseni (Bremer=3, bootstrap=68). Crown Alligator is less supported, but A. mississippiensis, A. mefferdi, and A. thomsoni for a very well supported clade (Bremer=5, bootstrap=90). Alligator thomsoni does not have any autapomorphies. The last clade is a well supported sister relationship between A. mississippiensis and A. mefferdi (Bremer=2, bootstrap=74). The two Asian species of Alligator included in this analysis were not returned as a single lineage. This was somewhat well supported; the bootstrap proportion was 55. It should perhaps not be ruled out entirely, as A. luicus is not well preserved, but there is more support for their being two separate lineages. Two putative alligatorines were recovered outside the clade. Chrysochampsa is returned as the basalmost globidontan. This is the first time it has been able to be included in an analysis, as previous attempts resulted in it being a wild card taxon a taxon recovered in so many positions that it caused the strict consensus to collapse (Brochu 1999b). A position outside Alligatoridae is not entirely unexpected, but being more basal than the Cretaceous globidontans was. Chrysochampsa is not well preserved.

30 16 Additional specimens and characters may cause it to be pulled crownward, but exactly where remains to be seen. I am highly skeptical of the position the Maoming gator was returned in as a basal caimanine. The only three characters causing it to unite with caimans are also present in Krabisuchus (no notch lateral to the naris and the lacrimal longer than the prefrontal) or exhibit homoplasy across the tree (upturned orbital margins). Characters that could potentially cause it to fall inside Alligatorinae were not visible. This taxon is even less well preserved than Chrysochampsa, with only its dorsal cranial surface exposed. Much of the superficial bone is prepped away, thus obscuring some sutures. The new partial vertebral column does not preserve many characters. Further prepping of both specimens is needed to expose the ventral surface of the cranium and the buried anteriormost cervicals. Excavation of the type locality is ongoing, so more specimens may be found in the future as well. For the purposes of my further analyses here, I still consider it a putative alligatorine and separately note any changes it would cause if it were sister to Krabisuchus, based on its geographic and temporal occurrence. It is worth noting that the Maoming alligator looks grossly similar to specimens of Hassiacosuchus that are also preserved with unsheared dorsal views of their skulls. While only conjecture at this point, it would not surprise me if it turned out to form a clade with Krabisuchus, the Ellesmere gator, and Hassiacosuchus. Geographically and temporally speaking, this would be in keeping with PETM dispersals to Asia across the Lomonosov Ridge and a dispersal to Europe across the Thulean land bridge.

31 17 CHAPTER II: BIOGEOGRAPHY Introduction Living alligatorids (alligators and caimans) lack adaptations that allow crocodylids to tolerate salt water (Jacobson 1983; Chen et al. 1989; Thorbjarnarson and Wang 2010). They lack lingual salt glands, a keratinous covering over their tongue, water resorbing glands in their cloaca, renal adaptations for processing high salt content, and their skin is less resistant to osmosis. All this combines to result in net water loss while in saltwater (see Grigg and Krishner, 2015 for review). Large alligatorines can tolerate brief contact with brackish or salt water but are not equipped for the sustained exposure necessary for intercontinental dispersal across vast bodies of water (Elsev 2005). When they do venture into saltwater, they keep their mouths closed in order to avoid the increased rate of osmosis that would occur were their mucosa to come in contact with saltwater (Birkhead 1981). Alligators are thus unable to eat when in water above some salinity threshold (Birkhead 1981). As such, rapid water loss when in high saline environments is thought to constrain alligatorine dispersal to non marine routes. The northern continents have a complicated tectonic history, resulting in the periodic rise and fall of land bridges and archipelagos between continents (Briakatis 2014). The continents themselves have also been in pieces in the past relative to their current configuration particularly Europe, which was an archipelago for much of alligatorine history. As such, alligatorine dispersal events are expected to correspond to the appearance and disappearance of terrestrial corridors. Climate is an additional limiting factor. As ectotherms, the high latitude dispersal pathways available to alligatorines would have been inaccessible when the climate was cold. Crocodylians have temperature dependent sex determination, with modern species optima for sex ratios from C occurring days after laying in the American alligator (Lang and Andrews 1994; Webb and Cooper Preston 1989). For more than a few hatchlings to be viable temperatures largely within the C range are needed (Lang and Andrews 1994). Individuals of all ages stop eating and go dormant when possible at 10 C (Grigg and Krishner, 2014; Thorbjarnarson and Wang, 2010)

32 18 Alligators are, however, the most cold tolerant of the living crocodylians, with many individuals able to survive short bursts of below freezing temperatures in winter by burrowing into dens or allowing themselves to be frozen under ice with only their nostrils exposed to the air (Brandt and Mazzotti 1990; Hagan et al. 1983, Brisbin 1982; Lee et al. 1997). The former behavior is favored by Chinese alligators, which dig long tunnels and hibernate in groups, though both extant species will dig dens for winter use (Thorbjarnarson and Wang 2010). The latter behavior, called icing, is more often done by A. mississippiensis, and includes a distinct posture that allows the animal to take advantage of the thermal gradient in water in order to stay warmer than they would if more of their body were exposed to air (Lee et al. 1997). Even hatchlings can successfully practice icing for very short periods of time, but they have much less room for error than larger individuals, who have more thermal inertia, and the former often die (Lee et al. 1997). Overall, cold winters are dangerous for hatchlings, but individuals 1.5 m or more in length frequently survive short bursts of these temperatures with no repercussions; 54 kg alligators were only 2 C warmer than 2 3 kg animals in one study (Seebacher et al. 2003). The northern limits of the American alligator s range corresponds with the isotherms for minimum January temperature = 9.4 C, mean minimum January temperature = 1 C, and January mean temperature = 7.2 (Neill 1971). One individual is known to have survived six winters in Pennsylvania when the mean air temperature during cold months was 5.5 before being shot at 1.25 m in length (Neill 1971). The biogeographic history of the Chinese alligator is particularly perplexing. Fossil first appearance dates and current phylogenetic hypotheses place the A. mississippiensis A. sinensis split in the early Miocene, when connections between North America and Asia were climatically unfavorable (Sun et al. 1992; Brochu 1999a). Molecular analyses using relaxed molecular clocks push this date back to Ma; these possess large error bars and would create multiple long range extensions, but the timing is more consistent with high latitude dispersal of an ectothermic freshwater lineage, as it was a much warmer time period (Wu et al. 2003; Oaks 2011). Multiple putative alligatorids have been reported

33 19 from Asia: Alligator luicus (Miocene; China), the Maoming alligatorid from Guangdong Province (Eocene; China), fossils referred to A. sinensis (Pleistocene; China, Taiwan, Thailand), Krabisuchus (Eocene; Thailand)(Sun et al. 1992; Martin and Lauprasert 2010a; Hsi yin et al. 2013), an unpublished specimen from Japan (Tanimoto, pers. comm.), Eoalligator chunyii (Paleocene, China), Eoalligator huiningensis (Eocene, China), and Lianghusuchus (Eocene, China), which is usually referred to as a crocodyloid, but may be an alligatoroid (Brochu 1999b). The relationships of these species to A. sinensis and one another are not resolved, as only Krabisuchus has been included in a phylogenetic analysis (Martin and Lauprasert 2010b). Inclusive analysis has the potential to revise fossil estimates of A. mississippiensis A. sinensis divergence time, and could reveal multiple dispersals to between North America and Asia, as suggested by Martin and Lauprasert (2010). Preliminary phylogenetic work suggested at least two dispersal events from North America to Europe during the Neogene (Brochu 1999a). This had been challenged by the discovery of alligator like fossils from the Late Cretaceous of Europe (Buscalioni et al. 1997; Martin and Buffetaut 2008; Delfino et al. 2008; Martin and Delfino, 2010), but these appear to be non crocodylians crocodyliformes (Turner and Brochu 2010; Brochu et al. 2011; Buscalioni et al., 1997; Martin 2007; Martin and Buffetaut 2008; Delfino et al 2008; Martin and Delfino 2010). A single dispersal event to Asia was expected for much of the time extinct alligatorines have been researched (Snyder 2007), but recent discoveries of Eocene alligatorines have suggested at least one more (Martin and Lauprasert, 2010; Skutschas 2014; Claude et al, 2011). Materials and Methods The Cenozoic paleogeographic history of the northern hemisphere is highly complex, with all three continents being connected at times and each continent sometimes in pieces. Normally the best program for a clade with restricted dispersal ability would be TreeFitter (Ronquist 1997; Sanmartin et al. 2008), but the reticulated relationships of continents and continent parts through time is not something TreeFitter can easily

34 20 accommodate, as it only accepts the input of nonreticulated cladograms of geographic relationships. Fortunately, Alligatorinae has a relatively simple biogeographic history, with nearly all its members being North American and a few isolated species on other continents. For the current analysis, I will use Lagrange for the quantitative analysis to reconstruct ancestral areas and qualitatively describe potential dispersal times and pathways. Lagrange is a process based method called Dispersal Extinction Cladogenesis (DEC) using a maximum likelihood model (Ree et al. 2005; Ree and Smith 2008; Nesbitt et al. 2009). It is able to incorporate complex temporal information and instantaneous rates of range transitions and is commonly employed by neontological workers. It allows for different likelihood values for dispersal between different areas, making it work well for clades like Alligatorinae, whose dispersal ability is inversely proportional to marine barrier width. I based my weights on: distance of water that needs to be traversed, if any; salinity of water in likely dispersal path; and known faunal exchanges. Specifically, the Bering, Thulean, and De Geer land bridges and subaerial Lomonosov Ridge; the Azolla blooms in the Arctic Ocean; and the series of mammalian faunal exchanges between Asia and North America in the early Miocene (later exchanges were ignored due to the increased aridity along the path); and the higher number of dispersals to Asia from North America than vice versa from Ma (Sanmartin et al. 2001). The only exception is that I maintained the weighting for dispersal between the Americas as 1 since I am not concerned with caimans in this study. I ran the analysis twice once with the decreased likelihood of Asia to North America dispersals in the Paleogene, and one allowing equal likelihoods going both ways (reported in the following format in the results: [equal biased]). I considered biogeographic reconstructions that fell outside of 2 log likelihood units of the maximum likelihood estimate to be suboptimal. I compiled all the information on the Lagrange configurator (Ree) and ran it in a Python Shell. Raw data and results files are in Appendix E. To incorporate the time between when the latest possible time clades could have split and the latest possible time terminal species could have arisen, I used the time between FADs

35 21 of nodes and species as branch lengths; in some cases branch lengths were 0. I used an MPT that minimized stratigraphic debt (A. mcgrewi closer to crown Alligator than A. prenasalis). While Lagrange requires ranges of two or more areas be possible for ancestral areas of clades, that does not make sense for this analysis given that a clade can arise in one location and one location only. I will discuss instances where combo areas are returned as most likely by discussing both included areas as separate possibilities. I used N for North America, E for Europe, and A for Asia. I did not break areas down further because it either did not matter for Alligatorinae (North America is a single continent throughout the Cenozoic), or because, in cases of multiple possible dispersal paths, there are either no fossiliferous sedimentary rocks of the appropriate age and environment preserved (Fennoscandia vs. the British Isles) or they have not been surveyed for alligatorine fossils (eastern Siberia vs. western Siberia). Results Lagrange returned the most likely ancestral area for Alligatorinae as N ( ). This was the case for most lineages, the exceptions being: A. luicus mississippiensis (NA ), A. luicus (A ), A. olseni mississippiensis (NA ), A. sinensis (A ). All older species outside North America were returned as dispersing from North America. Within Alligatorinae, the first two taxa to branch off the tree (Navajosuchus, and Ceratosuchus) are both North American. Navajosuchus is from the Paleocene, and Ceratosuchus from the latest Paleocene and earliest Eocene (Sullivan 1986; Bartels 1984). Another early but more derived alligatorine, Wannaganosuchus, is known from the late Paleocene of North America. Allognathosuchini is likely also North American in origin. While three of the four basalmost taxa in it are not North American (Krabisuchus is Asian and Hassiacosuchus and Arambourgia are European), all three have long ghost lineages. This unsampled record is consistent with the lineages having dispersed long distances from ranges originally at high latitudes in North America, where very few depositional facies are

36 22 preserved. Thus, the divergence of this lineage from Alligatorini would have taken place in the Paleocene, with dispersals to Asia (Krabisuchus) and Europe (Hassiacosuchus) sometime in the Late Paleocene to early (for the latter) to late (for the former) Eocene. The Maoming alligatorid likely dispersed in the early or middle Eocene as well, either as part of the same lineage as Krabisuchus, or as a separate dispersal if they are, in fact, unrelated. The Ellesmere gator (and fragmentary alligator material from Banks Island) should also be considered here. It is from the early Eocene (Estes and Hutchinson 1980) and its location would logically be along the dispersal paths to both Asia and Europe. The last European alligatorine was Arambourgia, in the late Eocene. It falls deep within the derived clade containing Allognathosuchus and Procaimanoidea, being more closely related to the latter. All other members of its subclade are North American. Procaimanoidea kayi is either early Eocene or middle Eocene (the age of its locality is not well constrained) (Self 2010), while P. utahensis is middle Eocene. Thus, Arambourgia likely split from Procaimanoidea in the early Eocene, but could have dispersed to Europe anywhere from the early Eocene to early late Eocene. The next youngest non North American alligatorine is Alligator luicus (Li and Wang 1987). The species up and down tree from it (A. mcgrewi and A. olseni, respectively) both appear in the early Neogene (the latter at the very beginning) (Schmidt 1941). A. prenasalis is the earliest diverging species prior to this nose, and is late Eocene in age (Mook 1932; Whiting 2014). No Oligocene alligatorine specimens are known (at least none identified to species level). Previous reports of A. prenasalis in the Oligocene are actually almost entirely Chadronian based on newer age constraints. So A. luicus would have split form other Alligator species sometime in the Oligocene and dispersed to Asia in either the Oligocene or early Miocene. The last non North American species to discuss are the extant Alligator sinensis and a cluster of fossils putatively assigned to it. Reports of A. sinensis from the Miocene are probably attributable to A. luicus, if they are actually alligators (teeth alone are unreliable for species level taxonomic assignment in nearly all crocodylians). Of the Pleistocene reports, the one from Taiwan (Shan et al. 2013) is morphologically consistent with A.

37 23 sinensis, but the skull from Thailand recorded in the literature as A. cf. sinensis looks very different from it (Skutschas et al. 2014). The odds are good that this skull represents a new species since like Krabisuchus and the Maoming alligatorid the pair are relatively close in time and space. The oldest close relative of A. sinensis is A. thomsoni, which is from the middle Miocene (Whiting 2014). There is also A. olseni, sister to the least inclusive clade containing A. sinensis and A. mississippiensis, which showed up at the beginning of the Miocene (Whiting 2014). An unpublished alligator from the Gray Fossil Site in Tennessese is probably somewhere in this part of the tree and is from the late Miocene (Schubert, pers. comm.). All this taken together means that the Chinese alligator likely split from its North American relatives no earlier than the early Miocene and dispersed at some later point in the Miocene. This is in disagreement with recent molecular divergence estimates, which place it at Ma (Oaks et al. 2011). Those would, however, result in multiple lengthy and unlikely range extensions within Alligator, and is poorly bounded, with only the two modern species of Alligator available to sample from, and their closest extant relatives, the caimans, splitting sometime around the K Pg boundary. Given the unlikely ghost lineages and the poor constraints, these molecular divergence dates, like other molecular divergence dates in the same situation, should be treated as the lower boundary for the divergence date, at best, much like FADs should be treated as upper boundaries. Discussion Lagrange The ambiguity of dispersals in Alligator was caused by A. olseni being surrounded by two Asian alligators. This is the only instance on the tree where immigration into North America was returned as likely. There was little difference in the likelihood between it and that of a North American origin (N ). Given the low resolution of the tree and the limited visibility of areas of interest in A. luicus, it is possible that A. luicus and A. sinensis actually form a lineage (along with the A. cf. sinensis from Thailand) and represent a single Neogene dispersal into Asia, but the possibility of two dispersals into

38 24 Asia currently has the most support. Because A. olseni appears in the rock record of midlatitude North America well before A. luicus and A. sinensis show up in Asia, dispersals to Asia, back to North America, then back to Asia are improbable. Climate Climate in the early Cenozoic was much warmer than it is now and quite amenable to alligatorine dispersal at high latitudes. An alligatorine occurred in a subtropical forest on Ellesmere Island (Estes and Hutchinson 1980). During the mid to late Paleocene, cm of rain fell there annually. There were several hyperthermals, with the Paleocene Eocene Thermal Maximum (PETM) being the most intense of them by far, with polar sea surface temperature increasing from C (Sluijs et al. 2006). During the middle Eocene, the sea surface temperature along the Lomonosov Ridge in the Arctic was 10 C from Ma and C immediately after (Brinkhuis et al. 2006). Temperatures around the Arctic Ocean were similar or slightly higher (Brinkhuis et al. 2006). On Axel Heiberg Island, adjacent to Ellesmere, data from plants pegs the middle Eocene mean annual temperature (MAT) at C, the cold month mean at 4 C, and the mean annual precipitation (MAP) at 122±37 cm per year. During the Oligocene, temperatures began to drop and precipitation in the northern hemisphere began to decrease. Ice rafted debris, evidence of the onset of glaciation, began to show up around 14 Ma, but was spotty until about 10 Ma (Via and Thomas 2006) There were two later warming periods in the late Oligocene and middle Miocene. The middle Miocene was warmer than the early Miocene or any time after, but high latitude dispersal during this period would have been much more difficult for alligatorines than before. Between the FAD of the smallest clade containing them and the slightly lower climate of the early Miocene, the late Oligocene is the most likely time for the dispersal of the lineage (or lineage) leading to A. luicus and A. sinensis, and would have occurred across Beringia. This is concurrent with the 23 Ma dispersal of mammals and precipitation would not have sharply decreased yet at this time (Webb 2006). While dispersal does not preclude prior cladogenesis, this is much younger than the sinensis mississippiensis divergence time estimated by molecular analyses (Oaks 2011), but as I

39 25 stated before, those dates should be treated as earliest possible times and other issues make them improbable. Dry Land Corridors Three land bridges were variously present during the history of Alligatorinae: Thulean, De Geer, and Beringia. Beringia is the most wellknown, and was present between Alaska and the Russian Far East. There are a number of hypotheses about the pattern of its emergence in the early Cenozoic, but multiclade floral and faunal dispersal data confirm definite subaerial exposure c. 65 Ma and c. 58 Ma (Briakatis 2014). One study on the biogeography of marine snails Figure 6: Map from Briakatis 2014 in the area supports its consistent emergence showing the location of the DeGeer (1), Thulean (2), and until the late Miocene (Marincovich 1999) Beringian (3) land bridges. The De Geer land bridge was present from northern Greenland, through Svalbard, and into northern Fennscandia and nearby Russia Ma (Briakatis 2014). Climatic evidence suggests dispersals across it occurred c. 69 Ma and 65.5 Ma (Briakatis 2014). The Thulean land bridge offered a route from southern Greenland through Iceland, the British Isles, and on into now continental Europe c. 57 Ma and 55.8 Ma (Briakatis 2014). Both of these corridors have the potential to be the migration pathway of Hassiacosuchus and Arambourgia. A DeGeer dispersal would require multiple long ghost lineages and is thus less likely than a Thulean dispersal. To more definitively pick between them, one would need to locate a sister taxon preserved along one or the other during the right time period. I have combed through many geologic maps and have unfortunately found no fossiliferous sedimentary rocks of the right age and environment that are currently subaerial. Glaciers have wiped away the Cenozoic rock record of Fennoscandia, Svalbard, Greenland, Iceland, and the northern British Isles.

40 26 A fourth land corridor which has only recently been discovered is the Lomonosov Ridge. Today it straddles the North Pole and is completely underwater, but during the PETM and early Eocene, it created either a solid corridor or a series of islands leading from Ellesmere Island and northern Greenland to the Russian Far East (Sluijs et al. 2008; Brinkhuis et al. 2006). Hopmans et al. created a method for measuring terrestrial organic matter originating in rivers to relative organic matter (the BIT index). Sluijs et al. (2006) applied this method to the Lomonosov Ridge core and concluded that it was subaerial during the PETM. The ridge gradually subsided during the early Eocene and was no longer present by the middle Eocene based on the BIT index, the relative scarcity of terrestrially derived palynomoprhs, and the geochemistry of clays (Sluijs et al. 2006, 2008; Brinkhuis et al. 2006). Unfortunately, this is the only deep sea core taken from the entire Arctic Ocean, much less from this location. As it only spans the uppermost Paleocene to middle Eocene (Brinkhuis et al. 2006), we currently have no way of knowing if the ridge was subaerial at any other time. Krabisuchus and the Maoming alligatorid could conceivably have crossed Beringia or the Lomonsov Ridge to get to East Asia. As with the European gators, finding timely relatives along dispersal paths in Alaska or Chukotka, Russia would be able to support one or the other. Relatives in Alaska or the northernmost reaches of North America would do the same. Most of these places have not been explored for alligatorines, but unlike for the European alligatorines, all of these places have formations which could potentially yield alligatorine fossils. The place that has been explored is northern Nunavut, and fragmentary alligatorine specimens have in fact been found on Ellesmere and Banks Islands. If, upon the discovery of more complete specimens, it is returned as sister to Krabisuchus, it would support a dispersal across the Lomonosov Ridge during or shortly after the PETM. This would also be in keeping with faunal exchanges between the eastern Nearctic and eastern Palearctic known to take place at this time (Sanmartin et al. 2001)

41 27 However, the Ellesmere gator is from the Bridgerian or latest Wasatchian. If it is sister to the Paleogene Asian alligatorines, this could mean that the dispersal happened later, perhaps during the Uintan, when many mammals dispersed (Woodburne and Swisher 1995), but a Beringian route is more likely in this case as such a great number of terrestrial mammals would not survive a lengthy swim in the ocean. But, as it turns out, an early Eocene swim across what was left of the subaerial Lomonosov Ridge cannot be discounted for an alligator. Salinity Large ectotherms are able to traverse long distances across bodies of water. The ancestors of New World Crocodylus crossed the open Atlantic (Brochu 2003). The saltwater crocodile regular swims between islands in the Indonesian archipelago (Grigg and Kirshner 2015). Alligatorines cannot due to their inability to tolerate saltwater. The Lomonosov Ridge core, along with a second core in the Norwegian Greenland Sea, also revealed multiple lines of evidence that the surficial waters of the late Paleocene and early Eocene Arctic Ocean was brackish, not salty, and at times even fresh (Brinkhuis et al. 2006; Sluijs et al. 2006). The Arctic Ocean has the lowest salinity of all the oceans today, verging on brackish in some locations. There are numerous lines of evidence that support an actual brackish state for its surface waters prior to the opening of the Bering Strait in the middle Miocene. These include isotopic analysis of fish bones (Waddell and Moore 2008), low diversity of marine dinoflagellates (Brinkhuis et al. 2006), sedimentological and geochemical evidence of increased stratification of Arctic waters from higher precipitation during the PETM (Dypvik et al. 2011; Harding et al. 2011) and through the middle Eocene (Stein et al. 2006), and models of Arctic Ocean Basin salinity which take tectonics, precipitation, and evaporation into account (Roberts et al. 2009).

42 28 An abundance of Azolla occurred in the middle Eocene for 700,000 years between 49 and 48 Ma (Brinkhuis et al. 2006; Sluijs et al. 2006). Azolla is a fern which floats on freshwater and cannot survive in briny or salty water, so its presence suggests an influx of freshwater which was not mixed with the saltwater already present in the Arctic Ocean basin (Brinkhuis et al. 2006). Similarly, cysts of crysophytes, a predominantly freshwater group of algae, co occurred with the Azolla, while the diversity of marine plankton dropped during these times (Brinkhuis et al. 2006; Sluijs et al. 2006). Their abundance was higher at the Lomonosov Ridge than in the Norwegian Greenland Sea, and occurred in thin cyclic laminae (Boulter 1986; Brinkhuis et al. 2006; Sluijs et al. 2006). These laminae are indicative of seasonal stratification of waters, with early spring being brackish, as determined by phytoplankton blooms, followed by a precipitation increase in late summer creating a freshwater surface which remained long enough for Azolla to bloom (Sluijs 2006). The lower abundance in the Norwegian Greenland Sea was a matter of slightly higher salinity due to the area s proximity to the North Atlantic. This gradient was potentially quite large, with analyses of ocean basin salinity returning 7 ppm off northern Greenland and 36 ppm off southern Greenland when a closed basin model was used (subaerial Beringia, which was at least partially present until the middle Miocene) (Roberts et al., 2009). Under an open model, northern Greenland waters were still briny (25 ppm), with southern Greenland waters at the border between briny and salty (30 ppm) (Roberts et al. 2009). Figure 7: Map from Brinkuis 2006 showing the location of the Lomonosov Ridge and documented Azolla blooms (stars).

43 29 An alternative reason for the presence of Azolla is that they were carried away from land by rivers. The lines of evidence that revealed a subaerial Lomonosov Ridge during the PETM also support a lack of terrestrial input during the middle Eocene when Azolla was abundant, so these ferns were in fact thriving on the open ocean near what used to be dry land (Sluijs 2006; Brinkhuis 2006; Sluijs et al. 2008). If any alligatorines dispersed during the early Eocene, they would have had an easier time swimming between islands along the way. If the ancestors of the Paleogene Asian alligatorines dispersed during the Azolla bloom, they could have island hopped across the last vestiges of the exposed Lomonosov Ridge. This possibility cannot be ruled out given the current state of our knowledge of the ridge s history. Further Arctic cores or paleontological exploration of Chukotka and Yakutia will be needed.

44 30 CHAPTER III: ECOMORPHOLOGY Introduction Alligator mississippiensis has a generalized diet, with individuals eating anything they can swallow. This includes insects, crustaceans, molluscs, fish, amphibians, reptiles, mammals, birds, fruit, and, more recently, marshmallows (Delany 1986, Brito 2002, Magnusson 1987, Taylor 1979, see Grigg and Krishner 2015 for review). The species is also said to have typical or average snout width and length (Langston 1973; Brochu 2001). This, in turn, has long been assumed to reflect the ancestral morphology for Crocodylia (Langston 1973; Brochu 2001). Preliminary phylogenetic work challenges this assumption. The earliest and most basal alligatorines were short snouted (brevirostrine) with enlarged blunt and bulbous cheek teeth specialized for durophagy that are not present in extant species (Brochu 1999, 2004)(FIG 8). Early brevirostrine alligatorines are superficially similar to the closest Figure 8: Globidont teeth in various taxa. Top row is: mosasaur, savannah monitor, relatives of Crocodylia (stem placedont. D-F are globidont teeth in fossil alligatorines. E,F show loss eusuchians), some of whom were in modern American alligator. previously misidentified as alligatorids (Brochu 1999, 2004). This warrants investigation, as it suggests that a living generalist is descended from a specialized ancestor. Rather than being plesiomorphic, A. mississippiensis may be long snouted (longirostrine) relative to its ancestral state. Another question would be how far back this plesiomorphic state extends. Alligatorines were diverse in the late Paleocene middle Eocene, when the clade was dominated by brevirostrine durophagous forms (von Othenio Abel 1928; Brochu 1999,

45 ). These specialized forms were mostly extinct by the end of the Eocene, with only the more generalized Alligator lineage persisting. Specialization is normally seen as an evolutionary dead end, with specialists at higher risk for going extinct as changing environments disrupt their narrow niches (Cope 1896; Futuyma and Moreno 1988; Ma and Levin 2006). Alligator may be a counterintuitive example of a specialist evolving into a generalist. This has been seen in parasitic arthropods adapting to a wider range of hosts (Johnson et al. 2009), but recognition of this reversal of the Law of the Unspecialized is still low. A study compiling data from neoecologic, paleoecologic, and phylogenetic work suggests that this is more common than we think, but exploration of this topic in the literature is still minimal (Colles et al. 2009). Here I explore the possibility of a reversal of Cope s Law of the Unspecialized and address a possible impetus for its occurrence. Materials and Methods My sample for the quantitative analyses consisted of 298 specimens in 81 species of eusuchians and stem neosuchians. I assessed ecomorphological trends in alligatorine evolution through traditional morphometric analysis using snout length plotted against size as a proxy. Traditional morphometrics is appropriate in this instance because onedimensional relative lengths are the trait being assessed, not more complex aspects of snout shape. Because a wide range of taxa has been included in this analysis, the measurement of size needed to be reliable across Eusuchia. After testing several other measurements, I chose the anterior width of the skull table as our size measurement (FIG 9). The skull table is a horizontally flat area overlying the braincase which is highly functionally constrained independent of snout length, and allowed a large sample of fossil taxa. Skull table length and posterior Figure 9: Snout length and size proxy measurements. width can vary slightly between crocodylian clades, but anterior width is more consistent.

46 32 A measurement of size used in previous studies (skull width at the posterior edge of the rostrum) has worked well in the single species analyses it was included in; however, these analyses did not include the range of snout morphologies I will be considering and thus cannot be decoupled from the functionality of the snout during feeding. A second size measurement published while this dissertation was being completed, width across the quadrates (Martin 2015), which are part of the jaw joint, is likely even more heavily influenced by feeding functionality, and should thus be discounted entirely for studies spanning a large and morphologically variable clade. Distorted and incomplete taxa were almost entirely excluded from this analysis. The exception being Procaimanoidea kayi, whose type and only specimen is missing the tip of its premaxilla. I included it because it retains an articulated dentary which allows snout length to be estimated with an error of only a few millimeters. I plotted relative snout length measurements in sequential order and used the asymptotes to quantitatively categorize each species into a snout length category. Cutoffs between longirostrine, mesorostrine, and brevirostrine were easily visible and correlated well with past qualitative assignments. There was a less noticeable asymptote approximately mid way through the mesorostrine category, so I used separate categories of shorter mesorostrine and longer mesorostrine to more finely quantify the continuum. There was also an asymptote separating the few most brevirostrine from others, but I have not considered it here since it may contain individuals verging on adulthood. I conducted ancestral state reconstruction in Mesquite using squared change parsimony, which is mathematically equivalent to a maximum likelihood model in which branch lengths are ignored (Maddison and Maddison 2007). For the ancestral state reconstruction, I selected large, adult individuals from each taxon and calculated the ratio of snout length to anterior skull width. A few fossil taxa are known only from small juveniles (e.g., Alligator luicus) and I did not include them in this analysis to avoid conflating ontogenetic and phylogenetic change.

47 33 I plotted the full sampling of snout length vs size to look for heterochrony in growth curves. Heterochrony is a shift in rate, starting point, or end point of growth trajectories resulting in a descendant looking more adult or more juvenile than its ancestor in the same ontogenetic stage. I will also examine the change in morphospace occupation in conjunction with the change in diversity through time. In addition to the quantitative analyses, I will qualitatively discuss other features known to be adaptations for durophagy, with the intent to also conduct ancestral state reconstruction for them in the future. Results American alligators have snout lengths almost equal to those of Crocodylus acutus in proportion to size, while tomistomines, gharials, and Mecistops are much more longirostrine. The long snout of Alligator mississippiensis reflects at least two forms of heterochrony: predisplacement, wherein the starting snout length is greater than in less derived forms; and hypermorphosis, wherein it continues growing beyond sizes reached by earlier forms and a relatively longer snout length follows since it always increases with ontogeny (FIG 10). It may also exhibit acceleration, wherein it has a faster growth rate (at least initially) than plesiomorphic forms. This is not clear, however, since the sampling of this region of the tree is sparse within each species and paraphylic groups of Figure 10: Growth trajectories in eusuchians and stem neosuchians. stem species understandably exhibit a wide range of size:snout ratios.

48 34 Other alligatorines, caimanines, stem globidontans, stem crocodyloids, and hylaeochampsids fall below the trajectory followed by A.mississippiensis and derived crocodyloids, quantitatively showing that, while A. mississippiensis has an average snout compared to modern Crocodylus, it is longirostrine relative to its ancestral condition. Alligatoroidea was recovered with an ancestral mesostrine snout (FIG 11). Chrysochampsa is an apparent (but not true) exception due to its abnormally narrow anterior skull table width. Otherwise, snout length began decreasing at Globidonta. Figure 11: Ancestral state reconstruction of snout length relative to size. Colors range continuously and indicate brevirostry (dark blue) to longirostry (red). Caimanines reversed this trend, but brevirostry persisted into Alligatorinae. In Alligatorini, it sequentially lengthened in derived forms, culminating in the longest alligatorine snout in A. mississippiensis. In contrast to this sequential lengthening, Allognathosuchini achieved extremely brevirostrine snouts matched only by Hylaeochampsids, Navajosuchus, Caiman latirostris, and Alligator cf. sinensis. The last is the most brevirostrine species (for its size) in this analysis.

49 35 Discussion Several cranial and dentary features are associated with durophagy. Enlarged back teeth may be bulbous and almost spherical, flattened, or rounded, but with a point or ridge on the apex (Crofts 2014, (Ősi 2011, 2014, Erickson 2012). These shapes maximize the ability of the animal to crack its prey open while minimizing strain and damage to themselves. Globidont posterior dentition is present in basal Alligatoroids and persists into Allognathosuchus, where the teeth are exceedingly large in all but some Willwood specimens. It is lost in the lineage leading to Procaimanoidea and in Alligatorini. The location in the back of the mouth allows the animal to apply more force without creating the undue stress on their jaws that would occur if bit down at the end of their rostrum (McHenry, 2006). American alligators can be observed applying this principle by shifting turtles to the backs of their mouths before applying enough pressure to crack their shells. Because the lengthy snout is not needed for a durophagous feeder, it follows that populations and taxa predominantly preying on hard prey might develop shorter snouts. In taxa which still engage in a large enough percentage of other, quicker prey, a longer snout would be retained to aid in prey capture. As demonstrated in the ancestral state reconstruction of relative snout length, Globidontans exhibit a progressive shortening from the snout length present in basal alligatoroids. Inside Globidonta, snout length increases in caimans and within Alligator. A progressive lengthening culminating in A. mississippiensis was reconstructed, but A. mcgrewi and A. sinensis both have shorter snouts than their closest relatives. A. luicus also has a short snout, but the only specimen is a juvenile, so whether or not this persisted into adulthood is unclear given the different snout growth curves exhibited by A. sinensis and A. mississippiensis. Additional modifications to the toothrow may include an enlarged anterior fang, a highly enarched region, and a series of small teeth in front of the back teeth (Jackson and Fritz 2004). These regions are not present on all durophagous taxa. In snakes that specialize on hard scaled skinks, these features aid in restraining frantic prey rather than

50 36 squeezing it out of the mouth as the jaw closes like a pair of scissors. This is not necessary in durophagous lizards that possess a form of cranial kinesis allowing them to bend their maxillae downward with a bite, much like the face Kermit the Frog makes when he s disgruntled. One example is the savannah monitor, which possesses only globular teeth that decrease slightly in size as one moves mesially. Gators have no cranial kinesis, so gaining these adaptations would be advantageous in highly durophagous taxa. These feature is, in fact, seen in them. A highly enarched dentary is one of the synapomorphies for Alligatorinae in this analysis. It is taken to an extreme in Allognathosuchus, and lost in Procaimanoidea + Arambourgia and Alligator. An enlarged 4 th dentary tooth shows a similar pattern, but is not lost in Alligator until after A. mcgrewi. Some Allognathosuchus also have many small, closely spaced teeth between the 4 th and 10 th dentary teeth, the latter of which becomes a second enlarged fang. Another feature notable in alligatoroids is their small adult size (FIG 12). The basalmost alligatoroids were comparable with large crocodyloids or, in the case of Deinosuchus, gigantic. In basal globidontans, size progressively decreases. Allognathosuchini are exceedingly small as adults, and the basalmost Alligatorinines are not much bigger. All these traits shifting together point toward an increased emphasis on durophagy by basal globidontans which is taken to an extreme in Allognathosuchus. The subclade containing Procaimanoidea and Arambourgia shifted away from this morphology and gained deeper snouts with more labiolinguially compressed back teeth. They may have been exhibiting more terrestrial ecology. A possible modern analogue is Paleosuchus, dwarf caimans, which hunt on land by lying in wait next to game trails at night, and thus do not need the flattened snouts that are advantageous for side to side head movement in water (Grigg and Krishner 2015). Alligatorini had very low diversity while Allognathosuchini was diversifying in the Paleogene and seems to have not been very successful until other alligatorines were gone, if the number of specimens preserved is

51 37 any indication, at which point they became larger and more longirostrine through progenesis and acceleration. Figure 12: Largest size sampled in each species as demonstrated by the anterior skull table width proxy. Red=gigantic, dark blue=very small. One possible reason for this drastic and sequential specialization for a durophagous niche (and, in the case of Procaimanoidea and Arambourgia, terrestrial ambush), is that they were pushed out of the large generalist niche early alligatoroids would have inhabited. I noted temporal and geographic co occurrences of other crocodylians and choristoderes with each species of alligatorine. In the Paleogene, brevirostrine alligatorines co occurred with large generalist crocodyloids (FIG 13). Crocodyloids were extirpated from mid latitudes by the end of the Paleogene and no longer co occur with alligatorines in the late Eocene. They are less cold tolerant than alligators and were thus unable to cope with the drop in temperature (Grigg and Krishner 2015). The two clades did not co occur again until very recently.

52 38 The only place alligators and crocodiles co occur today is in southernmost Florida. One pond famous with tourists contains both American species. When the crocodiles show up, the alligators, leave (Crocodile Specialist Group 2015 attendees pers. comm., Grigg and Krishner 2015). Crocodiles are much more aggressive than alligators, as evidenced by their higher propensity toward attacking humans, and by a video of ~6 ft American alligators being bullied into retreat by a pugnacious housecat who would not be alive if he lived near crocodiles instead of in Louisiana (Grigg and Krishner 2015, 10after ). The progressive lengthening in Alligator snouts started somewhat early in Alligatorini history and picked up in the late Neogene, until Alligator mississippiensis itself became long snouted enough to be longer mesorostrine verging on longirostrine. This morphology was already well established by the time Crocodylus dispersed into Florida. Their co occurrence is of little consequence to the species as a whole since they only overlap in a very small portion of the American alligator s range. Occurrence data and behavior inferred from modern species are in good agreement for niche competition being the reason globidontans were originally pushed out of large generalist morphospace and into more specialized niches which would not bring them into as much conflict with large crocodyloids. The Law of the Unspecialized is only a guideline, with the large generalist A. mississippiensis evolving back into that niche that was last inhabited by its ancestors in the Late Cretaceous. Spreading into an empty niche is a mechanism for specialists to evolve into generalists.

53 39 Figure 13: Co-occurrence of alligatorines and other crocodylian and choristodere species and their snout lengths.

54 40 CHAPTER IV: DIVERSITY DYNAMICS Introduction Alligatorines are the most cold tolerant of living crocodylians. They live in subtropical to warm temperate zones and are known to have ranged as far north as Ellesmere Island during the warmer Eocene (Estes and Hutchinson 1980). However, ectothermy, temperature dependent sex determination, and their subaquatic nature limit the range of climates a population can be sustained in (King and Dobbs 1975; Coulson et al. 1989; Brandt and Mazzotti 1990; Emshwiller and Gleeson 1997; Thorbjarnarson and Wang 2010). Because they live in their own depositional environment and possess dense bones and numerous osteological parts (osteoderms and constantly replaced teeth), they have a good fossil record and are less likely to be affected by rock record bias produced by unfavorable fossilization conditions than many other organisms. Climate sensitivity combined with a high chance of fossilization has made them useful as paleothermometers since their appearance in the fossil record 66 Ma (Markwick 1994, 1998b, a). Previous analyses indicate a bimodal diversity curve for crocodylians (including alligatorines) during the Cenozoic, with peaks in the early Paleogene and mid Miocene (Markwick 1994, 1998b, a). These correspond to global thermal maxima, suggesting a correlation between alligatorine diversification and climate. However, these did not incorporate phylogenetic information or more recently discovered species occurrences (largely Asian), and many of the included taxa had not been revised since the 19th century. Phylogenetic analyses can help reconstruct unobserved diversity by counting ghost lineages. Furthermore, we do not know to what extent observed drops in diversity reflect climate driven extinction or nonpreservation. Figure 14: Cenozoic climate change. Image source: de10.xml