Macroevolutionary drivers of extinction and

Macroevolutionary drivers of extinction and quantitative stratigraphy of graptolites

By James Boyle August 9th, 2018

A dissertation submitted to the Faculty of the Graduate School of The University at Buffalo, State University of New York in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Graduate Program in Evolution, Ecology, and Behavior

1

ACKNOWLEDGEMENTS AND DEDICATION I have an enormous number of people to thank for being able to finish my dissertation. First and foremost I’d like to thank my advisor Dr. Chuck Mitchell for being willing to support and encourage my work over not just the last five years of my dissertation but also during my master’s degree as well. It has truly been an amazing experience and I’m looking forward to continuing working with in the years ahead. Also, I feel I should apologize for the grammatical errors that are inevitably present in this text despite Chuck’s best efforts to root them out. I’d also like to thank the other members of my committee Dr. H. David Sheets and Dr. Scott Mackay for their guidance throughout this process. I have been extremely fortunate to work with Dr. Sheets who has a gift for explaining complex mathematical ideas in terms that are easy to understand. Perhaps most importantly Dr. Sheets helped me to accept uncertainty as a fact of our world and I feel I am better person for it. I’ve also been lucky to work with a few collaborators outside my committee who have offered their expertise over the last few years including Dr. Dan Goldman, Dr. Shuang-Ye Wu, and Dr. Mike Melchin. I’d also like to thank my lab mates who I’ve talked with over the years. Working through ideas, gathering samples in the field, or even just venting frustrations. Tayler, Rich, Erin, Michael, Steph, Brett, Patrick, and Shelby without all of you my time at UB would have been a very lonely place. Mentions of the lab would not be complete without thanking the small army of undergraduates who helped build up the biogeography database that makes up part of my dissertation. Special thanks to Nicole Leach, Rachel Kolenko, and Sophie Goliber as well as Nathan, Kristen, Pat, Gavin, Brian, Mike, Bilal, Phil, and all the others. I’d also like to thank all members of the evolution, ecology, and behavior program at UB for providing a rich inter-departmental experience during my time at UB. The mixing of ideas at seminars and the ability to take courses outside of my home department was a perfect fit for me and I hope that the program continues to grow and thrive in years to come. Special thanks to Sue, Sharon, Alison, Travis, and Robyn for keeping the Geology Department that was my home for seven years running smoothly. Finally, I’d like to thank and dedicate this work to my wife Alicia, son Seamus, and soonto-be baby girl. I made a choice to move back to Buffalo for graduate school to start a new life with my then-girlfriend and I’m so happy I did. Through the highs and lows of my graduate career Alicia has always been there to support me. I’ve become a more caring and understanding person because of her and I can’t imagine living my life with anyone else. I can’t wait to start the next chapter of our lives together.

2

TABLE OF CONTENTS LIST OF TABLES ….……………………..……………………………………………………4 LIST OF FIGURES …………………….…………….………………………………………...5 ABSTRACT …………………………….…………….…………………………………………8 PREFACE ………………………….…………………….……………………………………..11 CHAPTER 1: THE IMPACT OF GEOGRAPHIC RANGE, SAMPLING, ECOLOGY, AND TIME ON EXTINCTION RISK IN THE VOLATILE CLADE GRAPTOLOIDA……….…………………………………………………………………...….14 CHAPTER 2: QUANTIFYING GEOGRAPHIC RANGE MEASURES AND THEIR UTILITY AS CORRELATES OF EXTINCTION RISK …………………………………...66 CHAPTER 3: UNCERTAINTY IN HORIZON ANNEALING SOLUTIONS: A PRELIMINARY DARRIWILIAN ORDINAL COMPOSITE ……………………………112 FINAL REMARKS …………………………………………………………………………...140 REFERENCES ………………………………………………………………………………..144

3

LIST OF TABLES Table 1.1. (17) Factors correlated with extinction risk in the fossil record and their references. Table 1.2. (33) Sample size and description of each of the taxa subsets analyzed. Table 1.3. (37) Sample sizes, average and median durations, and standard deviations of durations for each run. Cells outlined in bold represent complementary sets of taxa. Table 1.4. (38) Wilcox tests for significant differences in geographic range and sampling measures between taxon subsets after false discovery rate correction. NS is p >= 0.05. * is p < 0.05. See Figure 1.4 for explanations of the column headings. Table 2.1. (81) Pearson correlation coefficients of the eBird (upper-right; n=1152) and PBDB (bottom-left; n=2348) geographic range measures and the number of unique locations taxa were observed at (NLoc). The PBDB correlation values are among cumulative geographic range measures of brachiopod genera observed in at least three unique locations throughout their duration. Table 2.2. (90) Spearman rank correlation coefficients and confidence interval boundaries between the Red List geographic range and geographic range values of 1152 bird species calculated from the eBird database. Table 2.3. (93) Results of general linear models of 381 brachiopod genera durations and geographic range measures. NLoc is the number of unique locations a taxon was observed at. Table 3.1. (118) HA parameters for searching solution space of the Darriwilian ordinal composite. Table 3.2. (126) Pearson correlation coefficients of the three measures of uncertainty across horizons. Table 3.3. (129) Uncertainty measures for the three Darriwilian solution trials and the summed uncertainty across all three trials for the vice and Island Search. Table 3.4. (129) Pearson correlation coefficient between uncertainty measures of horizons between trials.

4

LIST OF FIGURES Figure 1.1. (22) Distribution of graptolite biofacies groups and the inferred location of graptolite biotopes along a continental margin. Modified from Cooper and Sadler 2010. Figure 1.2. (28) A Mercator paleogeographic reconstruction of the Ordovician world at 460 Myr from PaleoGIS with the nine regions used in this study identified. Points on the map are occurrences of taxa examined in Boyle et al. 2014. Figure 1.3. (29) Map as in Figure 1.2 with five of the six gyres used in this study identified. Gyre 3 is placed between gyres 2, 4, and 5 during some 10 Myr time intervals. Figure 1.4. (34) Plot showing the correlation structure between sampling and geographic range measures used in the analyses. Arrows point to the factor with the highest correlation coefficient, which is labeled along the arrow. N.sect2 is the number of sites occupied counting composites as a single occurrence, n.loc is the number of distinct geographic locations occupied, n.reg is the number of paleoregions occupied, n.gyre is the number of paleogyres occupied, lat.rg is the latitudinal range in degrees, max.dist is the maximum pairwise distance, tr.sum in the summation of a minimum spanning tree of occurrences, n.bin is the number of 5x5° cells occupied, lon.rg is the longitudinal range in degrees, and area.ch is the convex hull area occupied. Figure 1.5. (41) Results from analysis of six different taxon sets, showing significant percentages of variance explained by single factors, PLSR axes, or GLM models of composite PLSR axes plus age cohorts and clade. Abbreviations as in Figure 4. Stars represent composite PLSR axes. Axis 1:n represent the summed variance in durations explained of the first to nth composite PLSR axes. Figure 1.6. (42) Results from analysis of seven different taxon sets, showing significant percentages of variance explained in durations by single factors, PLSR axes, or GLM models of composite PLSR axes plus age cohorts and clade. Abbreviations as in Figure 4. Stars represent composite axes. Axis 1:n represent the summed variance explained of the first to nth composite PLSR axes. Figure 1.7. (43) Stacked histogram showing the percent variance in durations explained by significant PLSR composite axes, age cohort, and clade for each of the thirteen taxon sets. Numbers below each bar are the significant percent variance explained by the best models, while the numbers above each bar represent sample size and total variance of durations. The ‘A’ and ‘C’ in the columns indicate (respectively) whether age cohort or clade alone, or explains the indicated variance in durations. Bold outline highlights the full set of 1114 taxa. Figure 1.8. (44) PLSR composite axis loading values for six of the thirteen taxon sets. Upward pointing black triangles represent positive loading (small solid: 0.2 – 0.6; large striped: > 0.6). Downward pointing gray triangles represent negative loading (small solid: [-0.2] – [-0.6]; large striped: