Applications of quantitative methods and chaos theory in Ichnology for analysis of invertebrate behavior and evolution

Trace fossils are the result of animal behaviors, such as burrowing and feeding, recorded in the rock record. Previous research has been mainly on the systematic description of trace fossils and their paleoenvironmental implications, not how animal behaviors have evolved. This study analyzes behavioral evolution using the quantification of a group of trace fossils, termed graphoglyptids. Graphoglyptids are deep marine trace fossils, typically found preserved as casts on the bottom of turbidite beds. The analytical techniques performed on the graphoglyptids include calculating fractal dimension, branching angles, and tortuosity, among other analyses, for each individual trace fossil and were performed on over 400 trace fossils, ranging from the Cambrian to the modem. These techniques were used to determine various behavioral activities of the trace makers, including feeding and behavioral evolution. Graphoglyptids have been previously identified as representing mining, grazing, farming, and/or trapping. By comparing graphoglyptids to known mining burrows and grazing trails, using fractal analysis, it was possible to rule out mining and grazing behaviors for graphoglyptids. To determine between farming and trapping, a review of all known trapping burrows was required. The hypothesis that graphoglyptids were trappers was based entirely on the hypothesized feeding behaviors of the worm Pciraonis. Close examination of Paraonis burrows indicated that the burrows are not traps. This means that, since Paraonis does not trap prey, graphoglyptids should not be considered traps either. Therefore, graphoglyptids likely represent farming behavior. This study also shows that previous interpretations of graphoglyptid behavioral evolution was far too simple. The results of the morphological analyses indicate that major changes to the behavioral evolution occurred during the Late Cretaceous and the Early Eocene. Previous hypotheses about Late Cretaceous evolutionary influences were validated. However there were additional influences like the Paleocene-Eocene Thermal Maximum that were not overly emphasized before. Finally, of the many theories about the driving force of evolution, chaos theory has often been overlooked. Chaos theory is a powerful tool, such that, by knowing the similarities between chaos theory and evolutionary theory, it may be possible to map out how environmental changes could shift the evolution of a species.

APPLICATIONS OF QUANTITATIVE METHODS AND CHAOS THEORY IN ICHNOLOGY FOR ANALYSIS OF INVERTEBRATE BEHAVIOR AND EVOLUTION by James Richard Woodson Lehane A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Geology Department of Geology and Geophysics The University of Utah August 2014 Copyright © James Richard Woodson Lehane 2014 All Rights Reserved The U n i v e r s i t y of Utah Gradua t e School STATEMENT OF DISSERTATION APPROVAL The dissertation of James Richard Woodson Lehane has been approved by the following supervisory committee members: Allan A. Ekdale , Chair May 5th, 2014 Date Approved Randall B. Irmis , Member June 6th, 2014 Date Approved Marjorie A. Chan , Member May 5th, 2014 Date Approved Elena A. Cherkaev , Member June 12th, 2014 Date Approved Leif Tapanila , Member June 6th, 2014 Date Approved and by John M. Bartley , Chair/Dean of the Department/College/School of Geology and Geophysics and by David B. Kieda, Dean of The Graduate School. ABSTRACT Trace fossils are the result of animal behaviors, such as burrowing and feeding, recorded in the rock record. Previous research has been mainly on the systematic description of trace fossils and their paleoenvironmental implications, not how animal behaviors have evolved. This study analyzes behavioral evolution using the quantification of a group of trace fossils, termed graphoglyptids. Graphoglyptids are deep marine trace fossils, typically found preserved as casts on the bottom of turbidite beds. The analytical techniques performed on the graphoglyptids include calculating fractal dimension, branching angles, and tortuosity, among other analyses, for each individual trace fossil and were performed on over 400 trace fossils, ranging from the Cambrian to the modem. These techniques were used to determine various behavioral activities of the trace makers, including feeding and behavioral evolution. Graphoglyptids have been previously identified as representing mining, grazing, farming, and/or trapping. By comparing graphoglyptids to known mining burrows and grazing trails, using fractal analysis, it was possible to rule out mining and grazing behaviors for graphoglyptids. To determine between farming and trapping, a review of all known trapping burrows was required. The hypothesis that graphoglyptids were trappers was based entirely on the hypothesized feeding behaviors of the worm Pciraonis. Close examination of Paraonis burrows indicated that the burrows are not traps. This means that, since Paraonis does not trap prey, graphoglyptids should not be considered traps either. Therefore, graphoglyptids likely represent farming behavior. This study also shows that previous interpretations of graphoglyptid behavioral evolution was far too simple. The results of the morphological analyses indicate that major changes to the behavioral evolution occurred during the Late Cretaceous and the Early Eocene. Previous hypotheses about Late Cretaceous evolutionary influences were validated. However there were additional influences like the Paleocene-Eocene Thermal Maximum that were not overly emphasized before. Finally, of the many theories about the driving force of evolution, chaos theory has often been overlooked. Chaos theory is a powerful tool, such that, by knowing the similarities between chaos theory and evolutionary theory, it may be possible to map out how environmental changes could shift the evolution of a species. iv This dissertation is dedicated to my lovely daughter, who has taught me the valuable lesson that no hardship is too tough, that you can't press on through it. "For our own species, evolution occurs mostly through our behavior. We innovate new behavior to adapt." -Michael Crichton, The Lost World (1995) TABLE OF CONTENT S ABSTRACT............................................................................................................. .....iii LIST OF TABLES........................................................................................................x ACKNOWLEDGEMENTS.................................................................................... ..... xi Chapters 1. INTRODUCTION AND BACKGROUND................................................... ..... 1 1.1 Overview..................................................................................................... 1 1.2 Graphoglyptids...................................................................................... .....4 1.3 Sampling..................................................................................................... 11 1.4 Significance of research....................................................................... ..... 20 1.5 Summary of following chapters................................................................22 2. FRACTAL ANALYSIS OF GRAPHOGLYPTID TRACE FOSSILS...................................................................................................27 2.1 Abstract....................................................................................................... 27 2.2 Introduction........................................................................................... ..... 28 2.3 Methods and materials............................................................................... 32 2.4 Results.........................................................................................................46 2.5 Discussion............................................................................................. ..... 47 2.6 Conclusion............................................................................................ .....51 3. PITFALLS, TRAPS, AND WEBS IN ICHNOLOGY: TRACES AND TRACE FOSSILS OF AN UNDERSTUDIED BEHAVIORAL STRATEGY....................................................................................................... .....52 3.1 Abstract....................................................................................................... 52 3.2 Introducti on........................................................................................... ..... 53 3.3 Modem trapping traces and their fossil equivalents.......................... .....54 3.4 Possible ancient traps................................................................................. 71 3.5 Discussion............................................................................................. ..... 77 3.6 Conclusion............................................................................................ .....79 4. ANALYTICAL TOOLS FOR QUANTIFYING THE MORPHOLOGY OF INVERTEBRATE TRACE FOSSILS...................................................... .....80 4.1 Abstract.......................................................................................................80 4.2 Introducti on........................................................................................... .....81 4.3 Methodology: Starting out................................................................... .....81 4.4 Morphology dependent methods...............................................................87 4.5 Morphology independent methods...........................................................101 4.6 Materials.....................................................................................................109 4.7 Discussion............................................................................................. .....I l l 5. BEHAVIORAL EVOLUTION OF BENTHIC ORGANISMS REFLECTED IN THE GEOLOGIC RECORD OF GRAPHOGLYPTID TRACE FOSSILS .............................................................119 5.1 Abstract.......................................................................................................119 5.2 Introducti on........................................................................................... .....120 5.3 Basis of analyses........................................................................................122 5.4 Ichnotaxonomy and topology groups.................................................. .....123 5.5 Materials.....................................................................................................134 5.6 Methodology......................................................................................... .....135 5.7 Results.........................................................................................................141 5.8 Discussion............................................................................................. .....163 5.9 Conclusion............................................................................................ .....177 6. EVOLUTION IN CHAOS: THEORETICAL APPROACH OF CHAOS THEORY AS A GUIDING PRINCIPLE FOR UNDERSTANDING PATTERNS IN BIOLOGICAL EVOLUTION.............................................. .....180 6.1 Abstract.......................................................................................................180 6.2 Introducti on........................................................................................... .....181 6.3 Chaos theory......................................................................................... .....182 6.4 Chaos theory in evolution.................................................................... .....185 6.5 Morphospace......................................................................................... .....188 6.6 B ehavi oral evoluti on..................................................................................191 6.7 Discussion............................................................................................. .....195 6.8 Implications........................................................................................... .....198 6.9 Conclusion............................................................................................ .....199 Appendices A. BENOIT PARAMETERS......................................................................................201 B. TORTUOSITY CALCULATIONS.................................................................. .....204 C. FRACTAL DIMENSION CALCULATIONS................................................. .....210 viii D. OCCUPIED SPACE PERCENTAGE AND BURROW SHAPE CALCULATIONS.................................................................212 E. SAMPLE IMAGES AND TRACES................................................................. .....216 F. LIST 01 TURBIDITE LOCALITIES............................................................... .....300 G. PALEOGEOGRAPHIC SAMPLE LOCALITY MAPS................................ .....310 H. DATA TABLES OF ANALYSIS RESULTS................................................. .....325 I. SAMPLE IDENTIFICATION AND ANALYTICAL PROCEDURAL ADJUSTMENTS................................................................... .....378 J. NETWORK TORUOSITY EXTENDED MATLAB SCRIPT........................ .....382 K. ANALYSIS CONSISTENCY TEST................................................................ .....398 L. BUILDING AN EXAMPLE EQUATION...................................................... .....406 REFERENCES........................................................................................................ .....412 ix LIST OF TABLES 2.1 Fractal analysis results of Zumaian trace fossils......................................... 48 3.1 Number of Paraonis burrows identified within the lower intertidal zone at Goose Point, Willapa Bay, Pacific County, Washington State at -40 studied locations................................... 69 3.2 List of possible trapping traces and structural features previously mentioned in the trace fossil literature...................................... 73 4.1 List of meandering and branching trace fossils and their values for various analytical methods for samples from Zumaia, Spain (Z) and Tanzania (T)............................................................ 112 4.2 List of meandering and branching trace fossils and their values for various analytical methods for samples from Zumaia, Spain (Z) and Tanzania (T) continued.......................................... 114 4.3 List of network trace fossils and their values for the various analytical methods for samples from Zumaia, Spain (Z) and Tanzania (T)........................................................................................... 116 5.1 List of graphoglyptids................................................................................... 134 6.1 Evolutionary factors that directly affect the evolution of a given species.................................................................................................. 198 F.l List of turbidite localities....................................................................................301 H. 1 Results of analyses on meandering trace fossils..............................................326 H.2 Results of analyses on spiraling trace fossils............................................. ......342 H.3 Results of analyses on branching trace fossils........................................... ......346 H.4 Results of analyses on network trace fossils.............................................. ......356 ACK N OWLED G EMEN TS This project, which has been five years in the making, has been helped and influenced by a lot of people and I would be remiss to not include or acknowledge at least a few of those people. I would like to thank my advisor, Tony Ekdale, for excellence in guidance and becoming an ichnological father figure to me. I thank the rest of my dissertation committee as well, Marjorie Chan, Leif Tapanila, Randy Irmis, and Elena Cherkaev, for providing guidance and insight into aspects of my research that I had not considered before and opening my eyes to new possibilities. I am especially indebted to Tommy Good, Andreas Wetzel, and Paula Dentzien-Dias for field work in Spain; Alfred Uchman, Jolanta Gruza, Quintin Sahratian, Elizabeth Gierlowski-Kordesch, and Mark Goodwin for museum, paleontological, and sample assistance; William Miller, Murray Gingras, Waldemar Obcowski, Andrew Milner, and Louis Buatois for geological and paleontological field and sample information; Dirk Knaust and Roy Plotnick for manuscript suggestions; Ron Bruhn, Lucas van Vliet, Kees van der Voort Maarschalk, Andreas Baucon, Jordi de Gibert, and Glen Mackie for mathematical and computer programming assistance; Malgorzata Bednarz and Serjoscha Evers for help in translations. Life in graduate school would not be complete without sharing the trials with other students, officemates, and friends. Those friends who have impacted this research in both big and small ways include Sherie Harding, Matthew Heumann, Jared Gooley, Ian Semple, Steve Pinta, Brendon Horton, Patrick Dooling, Luke Pettinga, Deanna Brandau, Morgan Rosenberg, Andrew McCauley, Tyler Szwarc, Joshua Lively, Carrie Levitt, Heather Judd, Mark Gorenc, Brenton Chetnik, Ryan Purcell, David Wheatley, Leah Toms, Julia Mulhem, Simon DeRuyscher, Nathan McClenathan, and Brett Nolan. While I have been here I have had the privilege to teach several labs under the guidance of John Bowman and Paul Jewell, without whom I would have been lost. Other funding was made possible by the ExxonMobil Geoscience Grant for Students, the William Lee Stokes Graduate Fellowship from the Department of Geology & Geophysics, and the Graduate Research Fellowship from the University of Utah. I also wish to acknowledge all of the societies and journals who have allowed me to reprint their material within this dissertation. Their individual acknowledgements are listed within the text. I would not be where I am today without the continued support through all of my schooling from my family, especially my mom, my dad, and my sisters. And last, but certainly not least, I could not have done this without the relentless support of my wife, Veronica, and the understanding from my daughter, Annabelle, who usually let daddy work when he needed to. CH A PT ER 1 INTRODUCTION AND BACKGROUND 1.1 Overview 1.1.1 Introduction Ichnology (the study of trace fossils) is an important field in geology and paleontology for many reasons, but mainly because trace fossils are autochthonous (found in place) indicators of paleoecological conditions. The autochthonous nature of trace fossils removes some of the doubt that is present when working with other paleoecological indicators (like invertebrate body fossils) that can become easily reworked, erasing important information in the process. Trace fossils provide an ecological usefulness that is not available from body fossils, because one animal will leave behind one skeleton (at most), but it could leave behind a countless number of footprints or a seemingly endless line of burrow trails, making them the most abundant type of fossils in the fossil record. Trace fossils result from animal behaviors, such as crawling, walking, burrowing, and feeding, which have been recorded in the rock record. For most of the nearly two centuries of ichnological research, focus has been mainly on the systematic description of trace fossils and their paleoenvironmental implications. Much less emphasis has been focused on how animal behaviors have evolved through time. 2 1.1.2 Purpose and approach The focus of this dissertation is to study the trace fossil record of invertebrate feeding patterns in the deep-sea through geologic time by employing quantitative descriptive methods. Behavioral evolution is an important topic, since animal behavior is one of the principal driving forces of evolution. Anatomical evolution proceeds in concert with behavioral evolution to drive the ways that species act and interact. The vast majority of studies in evolutionary paleontology deal strictly with morphologic changes expressed in body fossils. However, in exploring the long evolutionary history of life on Earth, the evolutionary trends of behavioral aspects should not be ignored. Without attempting to understand how behaviors evolve, we can only hope to understand one half of the equation. While some types of animal behavior are unpreservable, there are many aspects that in fact do have a potential for preservation as trace fossils. As discussed at length by many authors (e.g., see Seilacher, 1967, 2007; Ekdale et al., 1984a; Ekdale, 1985; Bromley, 1996; Mcllroy, 2004; Buatois and Mangano, 2011), these preservable aspects include diverse modes of feeding, dwelling, and locomotion. Ichnologists often describe trace fossils using vague descriptors like "narrow," "dumbbell-like," and "free meanders" (Hantzschel, 1975), but rarely, if ever, do they provide quantitative descriptions of the trace fossils. The problem with qualitative descriptions is that their precise meaning is hard to pin-point exactly. Different authors could assume different meanings for the same terms. Testing a qualitative hypothesis provides a weak test. For example, if something is either "meandering" or "not meandering," there is no middle ground with types, sizes, and/or degrees of meandering (Turchin, 1998). Quantitative descriptors offer a more precise basis for analyzing the 3 trace fossil. They are less ambiguous and subjective to what they are illustrating. A five centimeter wavelength on a meander is a five centimeter wavelength; there is no confusion. Qualitative hypotheses also allow for more definitive tests. Instead of a "yes" or "no" answer, there is "yes," "no," and "how much." If the previously mentioned trace fossil is meandering, there are specific questions that can be answered, like "how much is it meandering" and "are the sizes of the meanders consistent" (Turchin, 1998). One of the problems with Euclidean quantitative analyses is that there are a limited number of measurements that can be made. Beyond length, width, and thickness, ichnologists are often hard pressed to come up with other linear measurements that can be made. Lengths and widths also do not work well when dealing with trace fossils that could represent different ontogenetic stages of the trace maker's life or imperfect preservation of the trace fossil. When dealing with traces of varying ontogenetic stages, it is often best to use non-Euclidean (nonlinear) measurement techniques that will provide similar results for a wide range of trace fossils that contain a similar structure but vary in scale. In many cases, they produce consistent numbers that can be used as comparison tools among many different trace fossils. This approach also is useful for trace fossils that are incompletely preserved. Typically if you only have a fraction of a trace fossil, it is difficult to ascertain all the information you need from it, but with scale-invariant measurement techniques you can obtain similar results from a whole trace fossil as you would from a partial one. As long as the fossil is preserved as a complete, substantial piece, nonlinear techniques are useful. Computer simulations of idealized feeding patterns have been attempted by several workers (Raup and Seilacher, 1969; Papentin, 1973; Hammer, 1998; Plotnick, 4 2003; Plotnick and Koy, 2005), but computer analyses of actual trace fossils from the real world have not been accomplished to any appreciable extent. The analytical approach of this project includes morphometries (quantitative characterization of morphologic attributes) of actual graphoglyptid (discussed below) trace fossils in two-dimensional space using a variety of mathematical techniques, including fractal analysis (Mandelbrot, 1983; Feder, 1988; Slice, 1993), which has been applied only rarely in the field of ichnology (Jeong and Ekdale, 1996, 1997; de Gibert et al., 1999; Puche and Su, 2001; Le Comber et al., 2002; Romanach and Le Comber, 2004; Katrak et al., 2008; Baucon, 2010). Measurements of topology, tortuosity, occupied space percentage, branching angles, and burrow area shape also are employed. Use of such objective mathematical techniques allows a quantifiable interpretation of the trace fossils and provides a view of them with both size-dependent and size-independent parameters. 1.2 Graphoglyptids To study the evolution of behavior meaningfully, it is useful to limit the scope of the trace fossils that are being analyzed. The most promising trace fossils for analyses are those that are limited in sedimentologic extent and distributed over a long time period. The purpose of this project is to study the behavioral evolution of highly patterned deep-marine invertebrate feeding patterns, commonly referred to in the literature as "graphoglyptids" and/or "agrichnia" (Fuchs, 1895; Seilacher, 1977; Ekdale, 1980; Miller, 1991b; Uchman, 1995, 2003; Wetzel, 2000), using new quantitative methods (Fig. 1.1). The use graphoglyptid trace fossils enables the comparison of traces made in a very stable, consistent environment, since graphoglyptids are almost invariably preserved 5 Figure 1.1. Images of some common graphoglyptids. A) Cosmorhaphe from the Paleocene to Lower Eocene Variegated Shales of Poland. Sample number UJTF-2684. B) Helminthorhaphe from the Oligocene Krosno Beds of Poland. Sample number UJTF- 1362. C) Megagrcipton from the Early Eocene Guipuzcoan Flysch of Zumaia, Spain. Field photograph of sample labeled Z Megagraptonl. D) Pcileodictyon from the Jurassic Longobucco Sequence of Calabria, Italy. Sample number UUIC-721 E) Spirorhaphe from the Late Cretaceous of Kilwa, Tanzania. Sample number UUIC-1904. F) Urohelminthoida from the Messinian Azagador Limestone from the Vera Basin, Spain. Field photograph by A. A. Ekdale of a sample labeled VBUrohelml. Scale bars are 5cm. UJTF = Institute of Geological Science, Jagiellonian University, Krakow, Poland, Trace Fossil. UUIC = University of Utah Ichnology Collection, Salt Lake City, Utah. from the deep-sea on the base of turbidite beds, through time and across geographic boundaries. The deep sea is the largest and most stable single habitat on Earth. This fact suggests that animals and animal behaviors in the deep sea likely evolved very slowly through time (Seilacher, 1974). This dissertation provides insights about the rate of behavioral evolution, what factors influenced the evolution of behavior, and what feeding methods are represented by different burrow patterns. 6 1.2.1 Graphoglvptid history The study of animal behavior in the fossil record goes back at least as far as 1836, when Edward Hitchcock first started looking at vertebrate footprints preserved in Triassic strata of the Connecticut River Valley (Hitchcock, 1858). He recognized that the fossil trackways directly reflect the locomotory behavior of extinct beasts. The vertebrate traces were easier to identify, because ancient footprints preserved in rocks look very much like modern footprints in modem sand and mud. Invertebrate traces, however, were a little more difficult to discern. A century or two ago, many invertebrate traces were thought to be the products of algal growths (fucoids) or plants. Alfred Nathorst (1873) demonstrated that many of these types of fossils actually were organism traces that had counterparts in modern sedimentary environments. In 1895, Theodor Fuchs described a group of trace fossils that he found on the soles of turbidite beds. Fuchs used the term "Graphoglypten" to describe these problematical fossils that he noted as being raised reliefs on the underside of turbidite beds and often ornamental in design (Fig. 1.1). Fuchs (1895) noticed that although these graphoglvptid traces were diverse, they had enough in common to be considered a natural group of trace fossils. Fuchs initially suggested that these were casted surface tracks, but he subsequently concluded that they cannot be, because the original trace mold was never found, they never crossed one another, and there was no gradual coming and going of the tracks - they just appeared and disappeared. Fuchs' studies, along with those of other scientists, frequently lumped together graphoglyptids with flute casts and other basal turbidite features as "hieroglyphs" (e.g., Sujkowski, 1957; Dzulynski et al., 1959). A theory of formation that was presented by 7 Sujkowski (1957) for hieroglyphs was that they "are the infillings of impressions which were on the surface of the shale layer at the time of deposition." When working with turbidite deposits, Seilacher (1962) determined that there are two main types of trace fossils: predepositional and postdepositional. The predepositional trace fossils are those that are produced on the sea floor before a turbidite comes in and either destroys or preserves the upper layers of the sea-floor deposits. The post­depositional trace fossils are those produced in the turbidite deposits immediately or soon after the turbidite is deposited. Following a turbidite event, the postdepositional organisms produce the majority of traces until the food supply is exhausted and the bioceonosis returns to the normal quiescence of deep-sea life (Miller, 1991b). A few years later, Seilacher (1974) re-coined the term "graphoglyptid" for trace fossils that are "generally found on the soles of sandy flysch turbidites, as smooth and cylindrical casts" (Seilacher, 1977). He stated that they are open mud burrows that have been partially uncovered then casted by the overlying turbidite. With this terminology, he separated the hieroglyphs into two groups, the predepositional biogenic graphoglyptids and the nonbiogenic flute and cast structures produced by the turbidite. It was initially unclear whether graphoglyptids really represented open burrow systems, as had been hypothesized, or whether they were fecal-filled burrows where the fecal matter was stripped out during preservation. Ekdale (1980) put this issue to rest when modem graphoglyptids were discovered on the deep-sea floor in box cores. These observations showed conclusively that the graphoglyptids found on the soles of ancient turbidite beds were present in modern sediment as open tunnels. 8 1.2.2 Graphoglyptid preservation Modem graphoglyptid burrows are found mostly in the deep-sea. One exception of a trace that is sometimes grouped with graphoglyptids (Minter et al., 2006) is the burrow of the intertidal polychaete worm, Paraonis, which is discussed in Chapter 3. In the rock record, graphoglyptids are found on the base of turbidites, as was first described by Fuchs (1895). There are many hypotheses regarding how graphoglyptids originally were formed and how they eventually ended up as hyporeliefs on the soles of turbidite beds (Fig. 1.2). The current consensus is that the burrows started out as open tunnels, as seen in modern examples (Fig. 1.3 A). A turbidity current, also known as a density-driven gravity current, produced a mass of moving sediment intermixed with water that traveled along the sea-bottom. As the turbidity current moved down slope, it stripped away the surface veneer of sediment along its course. The sediment is removed from the seafloor by the suction power of the current front, pulling the sediment upwards into the water column, as opposed to scrapping it off of the surface as is typically assumed (Shanmugam, 1996). The open graphoglyptid burrows produce a weakened zone of sediment that allow the turbidite to remove the sediment from the top half, leaving the bottom half of the burrow intact. The sediment is incorporated within the turbidite and also kicked up into the water column. Closer to the more proximal limits of the turbidite, the amount of material that is stripped away is more significant, while further out from the proximal area, near the end lobes of the depositional fan, the amount of material removed is only a few millimeters off of the top of the bed (Fig. 1.3B). 9 Figure 1.2. Diagram illustrating the terminology for characterizing trace fossils depending on whether they are found on the top or bottom of the bed and whether they are either raised above or excavated into the bed. The largest particles in the turbidity current typically are sand-sized grains that settle in the bottom part of the open burrow, creating a cast of it (Fig. 1.3C). The cast that is preserved is a two-dimensional representation of a three-dimensional fossil. Most of the trace fossils that are analyzed in this dissertation are assumed to have been formed primarily in two-dimensional space (i.e., Paleodictyon, Spirorhaphe, Cosmorhaphe, etc.), but certain trace fossils were not analyzed, since they are assumed to be the remnants of a mostly three-dimensional trace fossil with only a cross-section preserved in two­dimensional space (i.e., Lorenzinia, Glockerichnus, etc.). Rock units that preserve extensive turbidite sequences are often known in European literature as "flysch" Even though this is primarily an archaic term, flysch is still in use in much of the current scientific literature. • • • • • • ' •W .V V.Relagie'Rairi '.V .V .V .* W . V W « VVV .V * V .* . VVV •• # t j m * ' • v > • • < ^ » • • v . •• ••* ••••• / J •• •v S •• r••%* •' •• t• • i r/ « ••• ••i i. •j • • • • • • • • • • • • • • • • • • • • • • • _ >_•_« • • ■ • B I • • (SteS^BtnasiJ530 Bottom View Figure 1.3. Diagram illustrating the preservation of graphoglyptid trace fossils. A) The open burrows of graphoglyptids are formed in the deep sea. B) A turbidite comes in and removes the upper layers of sediment overlying the burrows. C) Sandstone portion of the turbidite slab on which the graphoglyptids are preserved on the bottom. Rock unit shown in profile and bottom views highlighting the raised relief of the casted burrows. O 11 1.3 Sampling In order to investigate the broad scope of graphoglyptid trace fossil occurrences, a wide temporal and regional spectrum of sites needs to be examined. Examples for this project were chosen for their high abundance and diversity of graphoglyptid trace fossils in museum collections or at easily accessible field sites. Museum material is very useful, since a wide variety of trace fossils that have been collected previously can be photographed in a short period of time. In addition, pertinent field sites were visited, because not all trace fossil specimens are easily collectable. Some of the best preserved trace fossils occur in rocks that are too large to be collected and therefore must be left in the field. The four field sites that were studied in this project include the following (listed in geochronologic order): Point Saint George Turbidites, Franciscan Complex (Middle Jurassic to Middle Cretaceous), Northern California; Guipuzcoan Flysch, Higuer-Getaria Formation (Ypresian, Lower Eocene), Zumaia, Spain; Azagador limestone (Messinian, Miocene), Vera Basin, Almeria, Southeastern Spain; and intertidal deposits (Recent), Willapa Bay, Washington. Museum collections that were photographed include: University of Utah Ichnology Collection, Salt Lake City, UT; Institute of Geological Sciences, Jagiellonian University, Krakow, Poland; and the University Of California Museum Of Paleontology, Berkeley, CA. Most of the field samples were photographed in situ due to the constraints on sample collection and transport, both nationally and internationally, and large number of samples that were needed for the research. Samples for this dissertation were obtained in two ways. The first was taking photographs and samples from museum collections and in the field. Museum collections were photographed along with all pertinent information 12 including age, locality, and formation. A more complete analysis was supplemented by photographs from the literature representing graphoglyptids across the globe and geologic time. Uchman (2004) previously provided a comprehensive list of available graphoglyptid literature available. These were sorted and analyzed by ichnogenus, time period, and rock unit. The result was that the quantitative analyses were performed on more than 400 graphoglyptid specimens across the geologic time scale for this dissertation. 1.3.1 Field Sampling Localities 1.3.1.1 Point Saint George Turbidites, Franciscan Complex (Middle Jurassic to Middle Cretaceous), Northern California. This unit contains a rich graphoglyptid fauna not commonly seen in North America. Most of the turbidites represented on the west coast of the U.S. are slightly to heavily metamorphosed, so the presence of any trace fossils is not common and the presence of graphoglyptids in particular is rare. Graphoglyptids are preserved in the trench-slope basin or possibly the trench-floor setting (Miller, 1993). They include such distinctive graphoglyptid ichnotaxa as Belorhaphe, Megagrapton, and Squamodictyon. The turbidites here are considered to be inner- to mid­submarine fan deposits (Aalto, 1989). Due to the proximal location of the depocenter, this turbidite occurrence provides a different paleoenvironmental setting than most of the other turbidite examples to be studied for this project, where the deposition was more distal and sedimentation rate was lower. 1.3.1.2 Guipuzcoan Flysch, Higuer-Getaria Formation (Ypresian, Lower Eocene), Zumaia, Spain. Well-exposed sections of this graphoglyptid-rich turbidite 13 sequence are well-known for their high abundance and diversity of deep-marine trace fossils (Seilacher, 1977; Wetzel, 2000). Paleodictyon, Spirorhaphe, Cosmorhaphe and Helicolithus are especially abundant and widespread here. Samples and photographs were collected from Itzurun beach, near Zumaia, and up-section, approximately midway between Zumaia and Getaria (Fig. 1.4). Previous work has shown that this is a deep-water, siliclastic and calcareous turbidite interbedded with interturbidites and pelagic limestones. The facies that are represented by the turbidites are the basin-plain, outer fan, and deposition lobe of the middle fan (Leszczyiiski, 1991a), deposited on the order of one every few to several thousand years (Gawenda et al., 1999). The geological map used in this study (Fig. 1.4) is a combination of several geologic maps of the region, including those from Rosell et al. (1985), Pujalte et al. (2000), Bernaola et al. (2009), and Cummings and Hodgson (201 lb). These authors focused on different scales of the region and used different terminology for the rock units, hence the reason for an amalgamated map. The trace fossil-bearing units are in the Lower Eocene Higuer-Getaria Formation (also known as the "Eocene Flysch" [Bernaola et al., 2009] and the Jaizkibel Sequence [Rosell et al., 1985; Cummings and Hodgson, 201 lb]), the Lower Eocene Hondarribia Formation (also known as the "Eocene Flysch" [Bernaola et al., 2009]; and the Sarikola Sequence in Zumaia [Rosell et al., 1985; Cummings and Hodgson, 201 lb]), and the Upper Cretaceous Zumaia-Algorri Formation (also known as the San Telmo Red Carbonate Sequence [Rosell et al., 1985; Cummings and Hodgson, 201 lb]). 1.3.1.3. Azcigcidor Limestone (Messinian, Miocene), Vera Basin, Almeria, Southeastern Spain. The Vera Basin of Almeria in Southeastern Spain is unique among Quaternary deposits I Higuer-Guetaria Formation Hondarribia Formation Itzurun Formation Aitzgorri Limestone Formation I Zumaia-Algorri Formation Getaria France Zumaia Meters Figure 1.4. Geologic map of Zumaia, Spain, and surrounding region, showing sampling localities (A-C). The map is modified from Rosell et al. (1985, fig. 2), Pujalte et al. (2000, figs. 10 and 12), Bernaola et al. (2009, fig. 1), and Cummings and Hodgson (2011, fig. 1). ' 15 graphoglyptid localities, because it contains some of the youngest fossilized graphoglyptids in the world (Ekdale and de Gibert, 2014), with few to no other localities known that are geologically younger, except for the modem graphoglyptids (Ekdale, 1980). Another reason that this is a valuable graphoglyptid locality is that the environment of deposition is a relatively shallow (maybe only a few hundred meters deep), short-lived basin. The formations in which the graphoglyptids are found are hemipelagic marl deposits interbedded with turbidites that contain a diverse graphoglyptid ichnofauna (Braga et al., 2001). 1.3.1.4 Intertidal deposits (Recent), Willapa Bay, Washington. Ancient graphoglyptids have been related to modern burrows of the polychaete annelid, Paraonis, by some researchers (Roder, 1971; Seilacher, 1977; Minter et al., 2006). This small intertidal worm creates an open burrow system in a neatly spiraling pattern that rarely, if ever, intersects itself (Fig. 1.5). The geometric pattern and the open nature of the burrow system have made it a popular model for interpreting virtually all graphoglyptid feeding systems, especially the enigmatic double-spiraling graphoglyptid, Spirorhaphe. To evaluate the validity of this common claim, modem Paraonis burrows were analyzed. These are known to be accessible in intertidal flats of northern coastal regions, including the shores of Washington (Gingras et al., 1999), the Gulf of St. Lawrence (Brunei et al., 1998), and the North Sea (Roder, 1971). The closest locality to observe modern Paraonis in its burrow is in Willapa Bay, Washington (Fig. 1.6). This bay is a mesotidal estuary with a tidal range of 2 to 3 meters, and it is protected from the Pacific Ocean by the North Beach Peninsula. The sediments there are dominated by siliciclastic sand. Because Paraonis mainly occupies the middle to lower intertidal zone, Willapa Bay was visited 16 Figure 1.5. Horizontal view of Paraonis fidgens burrows at Goose Point, Willapa Bay, Pacific County, Washington State. Scale bar is 3 cm. during a spring tide when the tidal range was at its greatest (~3 meters). 1.3.2 Museum Sampling Localities 1.3.2.1 The University o f Utah Ichnology Collection, Salt Lake City, UT. The University of Utah Ichnology Collection (UUIC) in the Department of Geology and Geophysics houses more than 3,000 curated trace fossil specimens from all over the world and from multiple types of sedimentary deposits. The collection includes graphoglyptid specimens from North America, South America, Europe, and Africa, ranging in age from Jurassic to Miocene. 1.3.2.2 The Institute o f Geological Sciences, Jagiellonian University, Krakow, Poland. The Institute of Geological Sciences houses the extensive collection of 17 Figure 1.6. Location of modem Paraonis burrows at Goose Point, Willapa Bay, Pacific County, Washington State. Sampling locations indicated by the star. graphoglyptids amassed by M. Ksi^zkiewicz (1970) and other subsequent workers (e.g., Uchman, 1998). Samples come from many sites in Europe and range from Early Cretaceous to Oligocene. 1.3.2.3 The University o f California Museum o f Paleontology, Berkeley, CA. The University of California Museum of Paleontology houses the graphoglyptid specimens that were collected from the Point Saint George turbidites of Northern California. Samples are Early Cretaceous in age. 1.3.3 Scientific Questions and Hypotheses 1.3.3.1 Question 1. Deep-marine, preturbidite trace fossils, termed "graphoglyptids," are generally assumed to be geometrically complex and very regular in 18 shape. This observation suggests that it may be possible to quantify such shapes with a variety of methods ranging from fractal analysis to geometric morphometries. The question arises: Is it possible to quantify the fundamental shape attributes of graphoglyptids, and when the trace fossils are characterized quantitatively will they fit into ethologically meaningful categories of ichnogenera (Uchman, 2003)? In other words, can quantifiable geometric attributes of graphoglyptid burrows contribute to their systematic classification and to our interpretation of their paleoethologic significance? 1.3.3.2 Working Hypothesis 1. Graphoglyptid geometry tends to be sufficiently regular that similar trace fossils (i.e., ichnogenera) can be classified using quantification methods not normally used in the field of ichnology. Some types of burrow shapes are likely to produce a range of values, whereas other shapes will likely produce values diagnostic of that graphoglyptid ichnotaxon. 1.3.3.3 Onestion 2. The geometric regularity of the graphoglyptids presumably represents a distinct regularity in the feeding pattern of the organisms that created the traces. It has been hypothesized that the types of feeding behavior represented by graphoglyptids may be grazing (pascichnia), mining the sediment (fodinichnia), cultivating microbes inside their burrows (agrichnia), and/or trapping organisms passing through the sediment (irretichnia). It is reasonable to assume that some of these feeding patterns will yield a characteristic fractal dimension as well as other characteristic variables. The question arises: Is it possible that by analyzing the shapes of graphoglyptids, the feeding strategy can be discerned on the basis of quantitative measures? 19 1.3.3.4 Working Hypothesis 2. Some feeding patterns, like deposit feeding (grazing and mining), serve to optimize the amount of sediment that they utilize. Farming and trapping behaviors would be expected to have a more structured pattern, similar to a farmer's field and a spider web, respectively. These two groups of strategies (deposit feeding versus farming and trapping) would be expected to produce noticeably different morphometric results. Also, due to the wide variety of shapes and patterns of graphoglyptids, it is possible that there is a wide variety of feeding strategies that are represented by each specific graphoglyptid ichnogenus. 1.3.3.5 Onestion 3. In previous literature dealing with graphoglyptids, they often are described as perfectly formed nets, spirals, meanders, etc. (Seilacher, 1967, 1977; Ekdale et al., 1984a; Crimes and Crossley, 1991; Levin, 1994; Minter et al., 2006). The question arises: How closely do the actual trace fossils resemble the idealized perfect geometric forms that they represent? 1.3.3.6 Working Hypothesis 3. Graphoglyptids usually are regular in form, but on close inspection they are not geometrically perfect. Analyzing the graphoglyptid tunnel shapes allows us to illustrate this. It is expected that the patterns would be close to perfect, and any irregularities may be tied to specific environmental reasons (e.g., current direction, paleoslope, food concentration, etc.). 1.3.3.7 Onestion 4. Evolution typically is studied in the fossil record by looking at the changing anatomy of individual animals through geologic time. This is not possible for some kinds of animals, because they either do not leave behind a fossil record or have such a scant fossil record that it is not possible to study them from an evolutionary perspective. Trace fossils represent behavior, so by analyzing similar trace fossils, it 20 might be possible to determine the evolution of their behavior, by similar methods as paleontologists who study anatomical evolution. This has been accomplished for certain behavioral traits of modern organisms (Wenzel, 1992; Paterson et al., 1995; McLennan and Mattern, 2001; Price and Lanyon, 2002; Lopardo et al., 2004), but rarely has it been done for trace fossils. The question arises: Is it possible to study changes in behavior through time by just studying the trace fossils that have been left behind? 1.3.3.8 Working Hypothesis 4. It would be impossible to study the evolution of behavior by just looking at randomly selected trace fossils, but graphoglyptids are found in such a constrained environment with specific characterizations (deep marine; open burrow system; extremely shallow burrows) that it might be possible that the organisms producing the burrows are closely related and therefore that the behaviors they are illustrating can be linked in an evolutionary way. 1.4 Significance of research 1.4.1 Ichnologic significance The study of ichnology rarely delves into the quantitative realm, and when it does, it usually deals only with the percentage of disturbed sediment and the size of the trace fossils (e.g., Droser and Bottjer, 1986; Uchman, 2003). This dissertation expands the possibilities for studying trace fossils with scale-invariant (non-Euclidean) measures, such as fractal dimension and tortuosity, combined with more familiar Euclidean geometric parameters, such as branching angles of burrow tunnels. The methodology employed in this project enables ichnologists to study different varieties of trace fossils in a more objective manner than what is typically done. 21 1.4.2 Paleoecologic significance The functional purpose of graphoglyptids has long been debated (passive feeding, setting traps for other organisms, farming microbes, etc.), and the use of these quantitative methods helps to identify the feeding purposes of the trace makers. Different feeding methods are grouped on the basis of their quantitative attributes. By analyzing a variety of known feeding habit traces it is possible to interpret the feeding methods of graphoglyptids as well as provide tools for ichnologists to use for analyzing different trace fossils beyond graphoglyptids. 1.4.3 Behavioral evolution significance The fields of paleontology and behavioral biology seldom intersect. When they do, it typically involves trace fossils, since they are the tangible result of animal behavior, but behavioral evolution is rarely studied in the paleontologic record due to limited information. Some behavioral biologists have performed cladistic analyses on limited datasets of behavioral traits, and several workers have shown that with only behavioral characteristics it is possible to determine animal lineages (Wenzel, 1992; Paterson et al., 1995; McLennan and Mattern, 2001; Price and Lanyon, 2002; Lopardo et al., 2004). Studying the evolution of behavior in an ichnologic sense has been attempted only very rarely (Ekdale and Lamond, 2003), and in such analyses there generally are not the same level of mathematical standards that are now obligatory with modern cladistic and behavioral biology analyses. Even though this study is not cladistical in nature, it manages to bridge the gap between the two related fields of ichnology and behavioral biology. 22 1.4.4 Paleoclimate significance The deep sea is the largest, most stable habitat on Earth. Any changes to behavior in the deep sea should take a very long time unless something significant were to occur to alter the environment drastically. By studying the rate at which the graphoglyptid trace fossils changed it is possible to show paleoclimactic events which had a very large impact on the behaviors of organisms, and also to see if there were any large scale events, which possibly did not have any influence on the deep sea. 1.5 Summary of following chapters 1.5.1 Chapter 2 - Fractal analysis of graphoglyptid trace fossils The second chapter of this dissertation focuses on the feeding patterns represented by graphoglyptids, which have been interpreted as fodinichnial (mining), pascichnial (grazing), and/or agrichnial (farming). For this chapter, several species of graphoglyptid trace fossils were analyzed using fractal analysis to determine the fractal dimension of each of the traces. The fractal dimension combines shape complexity and space usage into one number. Fractal dimensions of graphoglyptid burrows were compared with those of known fodinichnial burrows, such as Zoophycos, and pascichnial trails, such as Scolicia, all from a similar time period and a consistent rock unit from Zumaia, Spain. The results from the study indicate that the deposit-feeding burrows (fodinichnia and pascichnia) illustrate a high fractal dimension. Graphoglyptids illustrate a consistently lower fractal dimension than the deposit-feeding burrows, thus providing evidence against the suggestion that they represent fodinichnial or pascichnial behaviors, supporting the hypothesis that graphoglyptids represent the agrichnial feeding habit. 23 1.5.2 Chapter 3 - Pitfalls, traps, and webs in ichnology: Traces and trace fossils of an understudied behavioral The third chapter reviews what types of behavior are contained within the group of agrichnial trace fossils. Previously, the term agrichnia has been applied to graphoglyptids and has been used to denote both trapping and farming behaviors, but the two behaviors display distinctly different feeding strategies. The trapping of prey is a specialized type of feeding behavior that is identified in the trace fossil record only rarely. Trapping traces that have been reported in the literature include spider webs, ant-lion burrows, scorpion pits, cerianthid tube anemone burrows, echiuran worm burrows, Paraonis worm burrows, and deep-sea graphoglyptids burrows. This chapter reviews all known trapping traces in both modern environments and fossilized occurrences. There is uncertainty, however, if all examples described as trapping traces truly represent traps. Paraonis burrows, for example, have been represented as trapping traces, but there is a question if they actually represent this kind of behavioral strategy. Previous references and new field work indicate that Paraonis likely employs a selective deposit feeding strategy. The interpretation that at least some graphoglyptids (e.g., Spirorhaphe) represent trapping was based on the trapping model for Paraonis, but since Paraonis does not trap prey, the question arises whether any graphoglyptids should be considered as representing trapping behavior. The new behavioral category ‘irretichnia' is proposed here to encompass trapping trace fossils, due to its unique behavioral significance and to separate trapping from farming. 24 1.5.3 Chapter 4 - Analytical tools for quantifying the morphology of invertebrate trace fossils The fourth chapter expands upon some of the analytical techniques that were introduced in the second chapter. The analytical techniques are meant to quantify the shape of trace fossils, enabling scientists to compare trace fossils described by different people with greater precision and accuracy. This chapter describes several methods for quantifying invertebrate trace fossils, including morphology dependent methods (motility index, mesh size, topology, tortuosity, branching angle, and the number of cell sides) and morphology independent methods (fractal analysis, burrow area shape, and occupied space percentage). These tools were performed on selected graphoglyptid trace fossils, demonstrating how these methods allow for objective comparisons between different trace fossils. 1.5.4 Chapter 5 - Behavioral evolution reflected in the geologic record of graphoglyptid trace fossils The fifth chapter addresses the evolution of graphoglyptid behaviors through time. The analytical techniques developed in the previous chapter were used on over 400 graphoglyptid traces that ranged in age from the Cambrian to the modern. Previous analyses of the behavioral evolution of graphoglyptids indicated that they were slowly diversifying, becoming optimized, and getting smaller over time until the Late Cretaceous, when a sudden increase in diversification occurred. This interval of rapid diversification of graphoglyptid ichnotaxa was likely attributed to the evolution of the angiosperms on land. Although some previous studies indicate that graphoglyptids were 25 getting smaller through time, results reported in this dissertation suggest that the feeding patterns they represent were not following any clearly established evolutionary trends. The behavioral evolution of the graphoglyptid trace makers was influenced many times during the past including the rapid diversification that started in the Late Cretaceous and continued through the Early Eocene, followed by a crash in diversity during the Oligocene. The initial diversity explosion was likely due to either the angiosperm evolution or an increase in foraminiferal/calcareous ooze and the Eocene diversification was likely continued because of the Paleocene-Eocene Thermal Maximum (PETM), which raised deep-sea water temperatures 4 to 5°C. The subsequent crash in the Oligocene was likely due to the Eocene-Oligocene Boundary Crisis which was a result of the growing ice sheets reducing sea level and increasing sediment deposition in the ocean due to erosion. Overall, graphoglyptids did not show the stability that is often attributed to them due to the stability and predictability of the deep-sea environment, but may in fact be sensitive indicators or deep-sea environmental change. 1.5.5 Chapter 6 - Evolution in chaos: Chaos theory as a guiding principle for patterns of anatomical and behavioral evolution The sixth chapter explores possible ways that chaos theory influences evolution. Previous applications of chaos theory in evolutionary studies have not taken the spotlight as the driving force of biological evolution. Chaos theory is based on nonlinear algebra, where the solution to one set of equations becomes the variable in the next iteration of the function, thus producing a feedback loop. There are six main principles to chaos theory, which can be related directly to biological evolutionary theory. Both theories embody 26 solutions to the problems that: 1, cannot repeat themselves; 2, are bounded within a specific region of space; 3, are heavily dependent on initial conditions; 4, are not random; 5, are unpredictable; and 6, are based on a series of feedback loops. Nonlinear systems can be depicted using a phase map, which illustrates all possible solutions of a problem depending on each initial value. In evolutionary theory, the phase map represents morphospace, which is the conceptual framework for mapping clusters of organisms based on specific attributes. The clustering of organisms is due to convergences, where many different genetic lines converge on similar solutions to various problems. The clustering in a phase map is concentrated around a point or region of space known as an ‘attractor'. External stimuli push the solution from the attractors to new attractors. Evolutionary external stimuli include changes in environmental factors, such as shifts in climate, or the introduction of new species. By using chaos theory as a template to study biological evolution, it may be possible to map out how human induced environmental changes could shift the evolution of species in the near future. CH A PT ER 2 FRACTAL ANALYSIS OF GRAPHOGLYPTID TRACE FOSSILS1 2.1 Abstract Graphoglyptids are a group of deep-sea trace fossils that exhibit ornate burrow geometries. Feeding patterns represented by these burrows have been interpreted as fodinichnial (mining), pascichnial (grazing), and/or agrichnial (farming). In this study, several different graphoglyptid trace fossils were analyzed quantitatively using fractal analysis to determine which of these three feeding modes is most appropriate as an interpretation. Graphoglyptid burrows lend themselves to fractal geometric analysis, because they commonly exhibit the essential fractal characteristics of scale invariance and self similarity. Fractal analysis is presented as a tool for analyzing geometric configurations by combining shape complexity and space usage into one number, the fractal dimension. Fractal dimensions of such graphoglyptid burrows as Pcileodictyon and Spirorhaphe were compared with those of known fodinichnial burrows, such as Zoophycos, and pascichnial trails, such as Scolicia, all from Zumaia, Spain. Results indicate that the deposit-feeding burrows (fodinichnia and pascichnia) illustrate a high Reprinted from Fractal analysis of graphoglyptid trace fossils, by James R. Lehane and A. A. Ekdale, PALAIOS, vol. 28, p. 23-32 with permission from SEPM (Society for Sedimentary Geology). 28 fractal dimension, as would be expected for a deposit feeding-optimal foraging strategy. Graphoglyptids illustrate a consistently lower fractal dimension than the deposit-feeding burrows, thus providing evidence against the suggestion that they represent fodinichnial or pascichnial behaviors. This observation supports the hypothesis that graphoglyptids represent agrichnial activity rather than mining or grazing activities. 2.2 Introduction Graphoglyptid trace fossils are geometrically complex, predepositional, open burrow systems commonly preserved in convex hyporelief on the soles of deep-sea turbidite beds (Fig. 2.1). The function of graphoglyptid burrows has been attributed to several different feeding strategies, including fodinichnial (mining), pascichnial (grazing), and/or agrichnial (farming) behavior patterns (e.g., Seilacher, 1974, 1977; Ekdale, 1980; Bromley, 1990; Rona et al., 2009). Fodinichnia (Seilacher, 1953), or sediment-mining traces, record the activity of an organism making repeated, closely spaced probes in the sediment to maximize the extraction of food resources. Pascichnia (Seilacher, 1953), or grazing traces, reflect the activity of a burrower feeding on organic material as the burrower moves through the sediment. Agrichnia (Ekdale et al., 1984a), or farming traces, are permanent (or semipermanent) dwelling burrows used for cultivating food. Seilacher (1974) suggested that graphoglyptids may be deep-sea feeding traces that developed geometrically complex patterns for efficiency of acquiring food, a strategy now sometimes referred to as optimal foraging. Optimal foraging strategy (OFS) refers to the maximization of the nutritional benefit from food versus the energetic cost of seeking 29 Figure 2.1. Images of some common graphoglyptids, all of which were photographed in the field in the Eocene Guipuzcoan Flysch of Zumaia, Spain. A) Cosmorhaphe. B) Helicolithus. C) Helminthorhaphe. D) Megagrapton. E) Paleodictyon. F) Spirorhaphe. Scale bars 4 cm. and exploiting a food resource (Charnov, 1976; Schneider, 1984; Plotnick and Koy, 2005). There are various applications of OFS, ranging from predator-prey relations to deposit feeding. A deposit-feeding optimal foraging strategy (DF-OFS) would apply where an organism ingests as much food-rich sediment as the organism can with as little effort as possible, in the process maximizing the coverage of the food-rich deposit (Levinton and Kelaher, 2004). The shapes of graphoglyptid burrow systems were thought to have become more geometrically complex throughout geologic time in a trend of increasing optimization during feeding (Seilacher, 1967, 1974, 1977, 1986). Alternatively, some workers have suggested that the graphoglyptid OFS was optimized early on but merely increased in geometric complexity for other poorly understood reasons (Crimes and Fedonkin, 1994; Uchman, 2003). Both of these suggestions seem to indicate that a DF-OFS was being employed by the trace-maker. 30 The agrichnial (farming) strategy is a behavior that seems to fit the geometric structure of graphoglyptids better than a DF-OFS strategy (Seilacher 1977, 2007; Rona et al., 2009). Support for the agrichnial hypothesis includes the possibility that graphoglyptids possess a mucus-lined wall, as might be suggested by the sharp outlines of the burrows when they are found as fossils. Geochemical tests of the burrow margin would be necessary to detect the presence of mucus. Rona et al. (2009) did such tests on modern Paleodictyon and found no evidence of mucus. The highly patterned graphoglyptid burrows likely represent a K-selected population strategy for survival in a stable, but resource-limited, environment where the burrowers have the time to build elaborate structures (Ekdale, 1985). The numerous openings of Paleodictyon to the sediment surface have been postulated as aeration holes, so that oxygenated water can be supplied throughout the burrow system. The main support for an agrichnial strategy is that most marine animals cannot break down the cellulose-based organic material that is found in the deep-sea environment (Seilacher, 1977, 2007; Cummings and Hodgson, 201 lb). For these animals to take advantage of the available organic matter, bacteria must be involved to breakdown the cellulose. The burrowers move back and forth through their burrow tunnels and consume the bacteria that they scrape off of the burrow walls. One of the questions arising from these hypotheses is whether or not all graphoglyptids are ethologically related, that is, if the unbranched burrows, branching burrows, and anastomosing burrow networks all represent the same activity. Some graphoglyptids may display one type of activity, whereas another group may display an entirely different activity. All graphoglyptids exhibit the same preservation mode (Fuchs, 31 1895), but this does not necessarily mean that all of the burrow patterns are functionally related or that the producing organisms are taxonomically related. There could be multiple evolutionary pathways to explain behaviors converging on a similar model of tunnel formation, but with completely different purposes. Each suggested graphoglyptid behavior (fodinichnia, pascichnia, and agrichnia) likely presents different quantifiable patterns in the rock record. The extent that the graphoglyptids exploit the sediment in which they are located could be an indication of the burrow's behavioral significance. Spiral shapes, meandering shapes, and networks are geometrically different, but if they cover the sediment in essentially the same way, the burrowers may be creating them for the same purpose (i.e., grazing or farming). The most promising way to compare the different types of burrow forms is to quantify their geometric configuration. Quantification offers an objective view of the behavior patterns that the shapes might represent. To study the geometric configuration of trace fossils requires a method that can give similar results for complete and incomplete trace fossils, as well as a method that will yield the same results at different scales (i.e., scale invariant). Fractal analysis is a useful method for expressing both the shape characteristics of the burrow and the extent of the sediment that is covered. The study described here tests the hypothesis that fractal analysis also can be meaningful in interpreting the type of behavior represented by the burrow geometry. Fractal analysis has been used in many animal behavior studies, including vertebrate foraging paths (Crist et al., 1992; With, 1994a, 1994b; Etzenhouser et al., 1998; With et al., 1999; Marell et al., 2002), vertebrate burrows (Le Comber et al., 2002; Romanach and Le Comber, 2004), invertebrate burrows 32 (Puche and Su, 2001; Katrak et al., 2008), simulated foraging paths (Plotnick,2003), and trace fossils (Jeong and Ekdale, 1996, 1997; de Gibert et al., 1999; Baucon, 2010). This paper is the first time that fractal analysis is used as a basis for describing different trace fossils in an attempt to interpret their behavioral significance. 2.3 Methods and materials Fractal geometry is a concept first described by Mandelbrot (1983) as a way to characterize complexity and quantify morphologies. The fractal approach can measure how completely a shape fills the space it occupies (Plotnick and Prestegaard, 1995; Wagle et al., 2005). A fractal is a shape that occupies a space where the precise dimension of that shape exceeds the topological dimension (Mandelbrot, 1983). A common view of fractals relates to their property of scale invariance, that is, when they look the same no matter the scale at which they are viewed (Fig. 2.2A). These are considered to be perfect fractals. A slightly looser, more pragmatic interpretation is that different scales of a fractal resemble the whole in some way (Feder, 1988). This definition applies to fractals that are not perfectly scale invariant but are considered to have a statistical scale invariance, i.e., a natural fractal (Slice, 1993; Plotnick and Prestegaard, 1995). This means that when the image is examined at different magnifications, all of the magnifications are not exactly the same; they just have a strong geometrical resemblance to one another (Fig. 2.2B). Fractals are illustrated by their fractal dimension (D), which is the space occupied as represented by a real number (allowing for a fractional dimension) rather than an integer. In Euclidian geometry, a straight line is one dimension, a plane is two 33 artificially created Koch curve (von Koch, 1904, 1993), where magnification of a tiny portion of the line results in exactly the same image as the previous view. B. A natural fractal. The naturally occurring Mississippi River drainage basin with a portion of the river drainage expanded to show its similarity to the whole. A scale bar is irrelevant in this figure because of the scale invariant nature of fractal images. dimensions, and a volume is three dimensions (Strogatz, 1994). When a shape takes up only a part of the space, the shape cannot be considered as occupying an integer dimension. A nonstraight line (e.g., a meandering trail) would occupy more than one dimension but less than two. An object radiating in the third dimension but not filling the third dimension (e.g., a subhorizontal branching burrow) would occupy a fractal dimension somewhere between two and three dimensions. The fractal dimension can be useful in understanding real shapes in nature, because the fractal dimension identifies the actual dimension that a particular shape occupies. Several methods have been used for calculating the fractal dimension. The fractal dimension for shapes in a two-dimensional space is best calculated using the Box Counting Method (D box) or the Information Dimension Method (Dinfo). D box 34 superimposes the shape on a grid (i.e., a set of boxes of specific size) and counts how many of the boxes are occupied by the shape in question (also see Baucon, 2010). The box size is then decreased by a set amount, and the boxes are counted again; this process is repeated over and over. The result yields a straight line in a log plot of box size versus number of occupied boxes, where the slope of the line is related to the fractal dimension (Fig. 2.3). Dinfb takes into account how much of the shape is located in each box, giving greater weight to the boxes with more of the shape inside the box. The application that was used to calculate the fractal dimension in this study was BENOIT Version 1.31 created by TruSoft Int'l, Inc., 1999 (Fig. 2.3; also see Appendix A). 2.3.1 Fractal properties of trace fossils There are two properties that a trace fossil should possess in order to be analyzed as a fractal. One property is scale invariance, where the image produces a similar fractal dimension no matter what scale at which the image is being observed. The other property is self similarity, where the image yields similar results when looking at different portions 2.3.1.1 Scale invariance. The problem with using fractal dimensions in ichnological studies is that very different ichnotaxa may illustrate a similar fractal dimension, while similar ichnotaxa may illustrate varying fractal dimensions. This even could apply to a single specimen, where different parts of the trace fossil may produce different fractal dimensions. Graphoglyptid burrows are so geometrically regular in comparison with many other types of trace fossils that this concern is expected to be minimal. 35 Figure 2.3. The fractal dimension program BENOIT, highlighting the location of the fractal dimension (1.475 in this instance) and the standard deviation (0.051 in this instance). The x-axis of the graph is the box size of each analysis, and the y-axis is the resulting number of boxes occupied. The fractal dimension is calculated from the slope of the graph. Also visible in this figure are the input parameters Side-length of largest box, Coefficient of box size decrease, Number of box sizes, and Increment of grid rotation (0­90). The removed analyses are highlighted (see Appendix A for description). Number of occupied box 37 To determine if a single image could produce varying fractal dimensions, a trace of a large (-25 cm wide) specimen of Paleodictyon majus (Meneghini in Peruzzi, 1880) from the Oligocene flysch of Poland was analyzed (Figs. 2.4-2.7). It should be kept in mind that all possible fractal dimensions (D) for forms that partially cover a plane are between 1.0 and 2.0. A dimension of 1.0 means the trace is a thin, straight line, and a dimension of 2.0 means that the trace completely covers the entire surface that is being analyzed. To test for scale invariance, the Paleodictyon majus specimen was analyzed at full scale, and then it was cropped down and rescaled in order to produce the same quality image for each scaled image (a-h in Fig. 2.5). This analysis was conducted using both D box and Dinfo, yielding results that ranged narrowly (D box from 1.450 to 1.553). In the graph (Fig. 2.5B), there is a reduction in the fractal dimension starting around 150 cm2, but the total reduction is only -0.09. Although this is a significant decrease on a narrow scale, when viewed in relation to the entire scale (1.0 - 2.0), this finding demonstrates a very good control (the average DBoxis 1.500 ± 0.058). Once the overall area of the Paleodictyon was reduced below -25 cm2, the results were appreciably different than the previous results, probably because at this point the form of the burrow stopped being a network and started to represent a branching form. 2.3.1.2 Self similarity. The Paleodictyon majus specimen then was analyzed for self similarity by taking the same size section (section "h" in Fig. 2.5) and finding the fractal dimension for various parts across the whole image (Fig. 2.6). The average fractal dimension (D box) of these boxes is 1.441 ± 0.023. Overall, the variation is far less in the self-similarity case than in the scale-invariance case. The D box range was 0.046 for the 38 Figure 2.4. Tracing of a Paleodictyon majus from the Oligocene Krosno Beds of Poland (un-numbered specimen in the Institute of Geological Science, Jagiellonian University, Krakow, Poland). A) Photo of Pcileodictyon majus. B) Tracing of the Paleodictyon majus specimen. Scale bar 4 cm. self-similarity test versus a range of 0.089 for the scale invariance test. 2.3.1.3 Eroding traces. There is an apparent relationship between the amount of preserved material in the sections in Figure 2.6 and the fractal dimension. More complete sections of the burrow (a, c, and d) have a higher fractal dimension than those where large portions are missing due to erosion (b, e, and f). This raises the question of whether an accurate (if slightly lower) representation of the fractal dimension is possible in situations where a trace fossil is incompletely preserved. To answer this question, the same Paleodictyon majus specimen was divided into 100 equal-sized sections (Fig. 2.7). A random number generator was used to remove (i.e., erode) parts of the trace until there was nothing left. This process was repeated ten times in order to demonstrate consistency of the results. The graph on the right side of Figure 2.7 shows what happens to the fractal dimension as the trace is eroded over ten iterations. The limits of the scale invariance and A B o B o xd im e n s io n • Information d im e n s io n Divide between network forms and branching forms 0.0 100.0 200.0 300.0 400.0 500.0 600.0 Area of the image (cm2) Scale invariance fractal analysis of Paleodictyon majus Figure 2.5. Fractal analysis of multiple scales of a Paleodictyon majus from the Krosno Beds of Poland. A) The fractal dimension (D) was calculated individually for each box after being scaled and reproduced using the same parameters listed in the text. B) Comparison of the Box Counting Method (D box) and the Information Dimension Method (Dmfb) to the area of the actual specimen in cm2. Error bars are the standard error that is calculated from the standard deviation calculated by BENOIT during the determination of the fractal dimension. Scale bar 4 cm, and error bars represent the standard error. A B Self similarity fractal analysis o f Paleodictyon majus 1.70 1.65 1.60 2cm 1.55 <D | 1.50 15 2 1 4 5 u. 1 40 1.35 1.30 O Box dimension □ Information dimension 1 4 - •- f - £ c d Section g Figure 2.6. Fractal analysis of multiple regions in a Paleodictyon majus from the Krosno Beds of Poland. A) The fractal dimension (D) was calculated for seven same-size sections (a-g). Section "a" corresponds with Section "h" on Figure 5. B) Comparison of the Dbox and the Dinfo for each of the sections in A. Scale bar 4 cm, and error bars represent the standard error. A B 1 4 X 5 6 7. V/ \ 10 11 r1 x2J> L '-' 13­16*/ H a ; \ 20 r 22- i r ? X -,25f - ^ . > ;28Cr?9'% 3 $ TWy -<32 23i 4it 13i, 5.' ,^ . i?V jt Y/ ov 41) > 4 ? rM - / ' 5!? !f53^ rx r \ rtV5i6 r y 4 rA 59 r Y(‘6'W3( ' y*u~ ••66' & 71 ^72 HA. 731;74 v ■ 75, y 76 A -71 v781 <:79r 80 81 82 83 84 *-85 86 QJ) 88 -8V9> ^90 r 91 92 93 94 95 96 97J -98; '100 Figure 2.7. Fractal analysis of Paleodictyon majns from the Krosno Beds of Poland as if the specimen were partially eroded. A) A random number generator was used to pick different portions of the Paleodictyon to be "eroded" and the resulting values were plotted after each 10% was removed. B) The plot of the eroded Paleodictyon calculated for ten iterations. The lines indicate the range of values of fractal dimensions reported in Figures 2.5 and 2.6. 42 self similarity analyses also are represented on this figure. The graph indicates that when the amount of the trace fossil that is eroded exceeds 40-50%, the values of D box and Dinfo begin to diverge, and they decrease below the values obtained in the previous two analyses. This erosion analysis demonstrates that even if as much as half of the original trace fossil has been removed by erosion, an accurate estimate of its fractal dimension is possible to obtain. One of the results that these three fractal analyses show is that the graphoglyptid burrow tunnels usually are thin enough that the percentage of area covered in each box does not give an appreciable difference in fractal dimension between D B o x a n d Dinfo. For this reason, only D box was used in the subsequent fractal analyses. 2.3.2 Fractals of feeding behavior One possible implication of using the fractal dimension is in the determination of a DF-OFS (deposit-feeding optimal foraging strategy) by graphoglyptids. With increased optimization, the burrowing pattern could become more geometrically complex and/or the burrowing organism could ingest more of the food-rich sediment. Either of these explanations would lead to a higher fractal dimension - most likely closer to D = 2.0 than to 1.0. It also is understood that several factors may affect how optimal or complex a forager's pathway could be, which could affect the fractal dimension of the resulting trace (Levinton and Kelaher, 2004). To test the hypothesis that optimization of DF-OFS leads to a higher fractal dimension, artificial burrow simulations of Papentin (1973) were analyzed. Papentin (1973) developed a computer program where a society of approximately 100 virtual 43 "worms" (nicknamed "Rectangulus rectus ") were allowed to "evolve" following a set of parameters derived from an analysis of burrows of the modem polychaete worm Paraonis. Using the images that Papentin ( 1 9 7 3 ) published of a few selected steps, the fractal dimension was calculated for each step and then plotted (Fig. 2 .8 ; Papentin, 1 9 7 3 , fig. 4 ) . Over the course of 1 4 0 generations for a population of approximately 1 0 0 "worms," the fractal dimension increased each subsequent time from D box = 1 .6 5 9 to 1 .7 5 2 . This analysis shows that the optimization of the feeding patterns occurred very quickly at first and then proceeded much more slowly after the sixth generation. 2.3.3 Comparing trace fossils The Rectangulus experiment of Papentin (1973) represents the development of one type of systematic feeding pattern in the sediment, as was discussed for the DF-OFS. The most instructive way to determine what ethologic strategies might have been used by graphoglyptid producers would be to compare real graphoglyptid specimens with other types of trace fossils in a similar time period and environment. For this study, a number of graphoglyptid examples from the Guipuzcoan Flysch of the Higuer-Getaria Formation (Lower Eocene), Zumaia, Spain were examined (Fig. 2.9). Well-exposed sections of graphoglyptid-rich turbidite sequences in this region are celebrated for their high abundance and diversity of deep-marine trace fossils (Seilacher, 1977; Wetzel, 2000). The turbidite facies represented include basin plain, outer fan, and depositional lobes of the middle fan (Leszczynski, 1991a), where each turbidite layer was deposited once every few thousand years (Gawenda et al., 1999). The stratigraphic sequence contains an uninterrupted succession from the Upper Cretaceous through the I Generation 6 if- ini linn Hint. Generation 30 lllllllll IHI II IIIIHIIII Generation 50 1.80 1.78 1.76 1.74 1 1 72 cd) ■E5 1.70 3o 1.68 2 1.66 1.64 1.62 1.60 ,H l| Generation 140, £ m in ^ iiif a mm I ! Fractal analysis of Papentin's (1973) Rectangulus rectus artificial evolutionary experiment 20 40 60 80 Generation 100 120 140 160 Figure 2.8. Papentin's (1973, fig. 4) computer-generated evolutionary experiment. A) Generation 0. B) Generation 2. C) Generation 6. D ) Generation 30. E) Generation 50. F) Generation 140. G) The plot shows a comparison of generation versus fractal dimension (D ) for the D box. Error bars represent the standard error. I Quaternary deposits I Higoer-Guetana Formation I Hondambta Formation | Itzurun Formation j Ait2gorri Limestone Formation : Zumaia-Algorri Formation Getaria Zumaia Figure 2.9. Geologic map of Zumaia, Spain, and surrounding region, showing sampling localities (A-C). The map is modified from Rosell et al. (1985, fig. 2), Pujalte et al. (2000, figs. 10 and 12), Bernaola et al. (2009, fig. 1), and Cummings and Hodgson (2011, fig. 1). L4^/i 46 lower Eocene. Trace fossils were photographed in the field for analysis, since the rock slabs were too large to be collected. The graphoglyptid specimens that were analyzed occur at a single locality along the coast between Zumaia and Getaria (Point C on Fig. 2.9). The representative graphoglyptids include Cosmorhaphe, Helicolithus, Helminthorhaphe, Megagrapton, Spirorhaphe, and Paleodictyon (Fig. 2.1). All of these graphoglyptids are understood to represent burrowing activity in a 2-dimensional space. Examples of Lorenzinia and similar radiating traces were not analyzed, since they likely represent only a portion of a 3-dimensional trace. The graphoglyptids were compared to known deposit feeding traces, including both pascichnia (Scolicia) and fodinichnia (Zoophycos), which were selected from within the same sequence of rocks. Zoophycos samples are Late Cretaceous in age (Point A on Fig. 2.9), and Scolicia samples are Early Eocene in age (Points B and C on Fig. 2.9). 2.4 Results The fractal dimensions were determined for all of the analyzed trace fossils (Fig. 2.10; Table 2.1). Results show that the fractal dimension of sediment deposit-feeders is significantly higher than that of any of the graphoglyptids. The D box of the sediment deposit-feeders ranged from 1.767 to 1.850, whereas the Dbox of the graphoglyptids ranged from 1.277 to 1.626. In the case of "Rectanguhis" and Scolicia, the reason for the higher Dbox is because there are multiple pathways that overlap each other. This likely resulted from the same organism (or multiple organisms) retreading the same ground. In the case of Zoophycos, the reason for the higher DBoxis because this is an example of a 47 Fractal dimension of selected trace fossils from Zumaia 2.00 1.90 1.80 1.70 C0 ■| 160 o> 1 150 ro t> 1 40 03 1.30 1.20 1.10 1 00 ! t i i □ Ichnoscncra Figure 2 .1 0 . Fractal dimension of selected trace fossils from Zumaia, Spain. The fractal dimension (D box) scale extends for the entire range possible for planar trace fossils ( 1 .0 - 2 .0 ) . Ichnogenera include graphoglyptids (Cosmorhaphe, Helicolithus, Helminthorhaphe, Megagrapton, Paleodictyon, and Spirorhaphe) and nongraphoglyptids (Scolicia and Zoophycos). Individual sample values are listed in Table 2 .1 . Error bars represent the standard error. The asterisk indicates a sample that was significantly eroded. single organism being extremely efficient and producing very tightly packed feeding tunnels. 2.5 Discussion The use of fractal dimensions to analyze various trace fossils supports the theory that graphoglyptids represent agrichnia instead of a deposit feeding strategy (pascichnia Table 2.1. Fractal analysis results of Zumaian trace fossils Trace Sample D box Trace Sample D box Trace Sample D box Cosmorhaphe 1 1.405 ±0.003 Megagrapton 5 1.491 ±0.005 Spirorhaphe 4 1.626 ±0.006 Cosmorhaphe 2 1.482 ±0.026 Paleodictyon 1 1.468 ±0.017 Spirorhaphe 5 1.604 ±0.005 Cosmorhaphe 3 1.373 ±0.011 Paleodictyon 2 1.588 ±0.011 Scolicia 1 1.850 ±0.002 Cosmorhaphe 5 1.456 ±0.028 Paleodictyon 3 1.507 ±0.013 Scolicia 2 1.848 ±0.002 Cosmorhaphe 6 1.382 ±0.009 Paleodictyon 4 1.507 ±0.011 Scolicia 3 1.802 ±0.003 Helicolithus 1 1.377 ±0.010 Paleodictyon 5 1.478 ±0.008 Scolicia 4 1.777 ±0.007 Helicolithus 2 1.277 ±0.008 Megagrapton 1 1.423 ±0.010 Scolicia 5 1.827 ±0.006 Helicolithus 3 1.268 ±0.012 Megagrapton 2 1.457 ±0.020 Scolicia 6 1.767 ±0.003 Helicolithus 4 1.365 ±0.002 Megagrapton 3 1.344 ±0.011 Zoophycos 1 1.811 ± 0.002 Helicolithus 5 1.413 ±0.023 Megagrapton 4 1.399 ±0.011 Zoophycos 2 1.830 ±0.001 Helminthorhaphe 1 1.489 ±0.002 Megagrapton 5 1.491 ±0.005 Zoophycos 3 1.804 ±0.004 Helminthorhaphe 2 1.498 ±0.005 Paleodictyon 1 1.468 ±0.017 Zoophycos 4 1.807 ±0.002 Helminthorhaphe 3 1.528 ±0.022 Paleodictyon 2 1.588 ±0.011 Zoophycos 5 1.776 ±0.002 Helminthorhaphe 4 1.468 ±0.018 Paleodictyon 3 1.507 ±0.013 Zoophycos 6 1.797 ±0.001 Helminthorhaphe 5 1.566 ±0.005 Paleodictyon 4 1.507 ±0.011 Zoophycos 7 1.774 ±0.001 Megagrapton 1 1.423 ±0.010 Paleodictyon 5 1.478 ±0.008 Zoophycos 8 1.766 ±0.002 Megagrapton 2 1.457 ±0.020 Spirorhaphe 1 1.573 ±0.004 Zoophycos 9 1.821 ±0.001 Megagrapton 3 1.344 ±0.011 Spirorhaphe 2 1.571 ±0.006 Zoophycos 10 1.787 ±0.002 Megagrapton 4 1.399 ±0.011 Spirorhaphe 3 1.419 ±0.009 The fractal dimension of selected trace fossils from Zumaia, Spain, that are illustrated in Figure 2.10. Ichnogenera include graphoglyptids (Cosmorhaphe, Helicolithus, Helminthorhaphe, Megagrapton, Paleodictyon, and Spirorhaphe) and non-graphoglyptids (Scolicia and Zoophycos). Variance in the fractal dimensions is represented by the standard error. 49 and fodinichnia). The producers of graphoglyptids were not deposit feeders, based on comparisons of the fractal dimensions of the graphoglyptid burrows with those of known deposit-feeding (pascichnial and fodinichnial) traces. In this study, a systematic deposit-feeding strategy creates a higher fractal dimension (in this study, greater than 1.75; see Table 2.1). The higher fractal dimension is possibly a result of the organism feeding from as much of the sediment as possible. The fractal dimensions of both Scolicia and Zoophycos are more representative of complete coverage of the sediment area (D box = 1.767 to 1.850) than what is seen for any of the graphoglyptids (D box = 1.277 to 1.626). This finding suggests that the tightly spaced geometric attributes of deposit-feeding burrows, which yield a high fractal dimension, is a result of a feeding pattern that maximizes the coverage o f a food-rich path of the sediment. The fractal dimension of the graphoglyptid burrows is too low compared with that of the deposit-feeding burrows in the same geologic age and setting, although repeated tests would be needed to confirm these results. Jeong and Ekdale (1996, 1997) suggested that the fractal dimensions of some Paleozoic deposit-feeding burrows reflected the efficiency of systematic feeding within the sediment. De Gibert et al. (1999) used fractal analysis to analyze the geometry of the trace fossil Sinusichnus. The fractal dimension of Sinusichnus was determined to range from 1.22-1.58, corresponding with the fractal dimension range of graphoglyptids in this study. The low to medium fractal dimension results were used to support the hypothesis that Sinusichnus represented an agrichnia trace. Baucon (2010) calculated the fractal dimensions of various types of trace fossils, including some graphoglyptids, from specimens illustrated in the literature. According to his results (Baucon, 2010, fig. 14), 50 the highest fractal dimensions are seen in fodinichnia (Zoophycos) and pascichnia (Helminthoida), whereas the fractal dimensions of graphoglyptids (Paleodictyon, Spirorhaphe, Cosmorhaphe) are lower. Baucon's (2010) calculations conform well to those obtained in the current study, and they corroborate the observation that graphoglyptids typically display a noticeably lower fractal dimension than what might be expected for a fodinichnial or pascichnial feeding strategy. Examination of the fractal dimensions of the different graphoglyptid ichnogenera analyzed for this study show that most of the fractal dimensions overlap significantly. This is strong support of the hypothesis that the different ichnogenera were using the same behavior, since similar fractal dimensions indicate that they covered the surface to the same extent, even though they are composed of significantly different shapes. These interpretations lead to the conclusion that graphoglyptid burrows likely represent an agrichnial strategy. The overlap of fractal dimensions also lends credence to the hypothesis that graphoglyptids comprise a single ethologic group of trace fossils and not an amalgamation of similarly preserved trace fossils. Results of the analyses in this study (Fig. 2.10; Table 2.1) indicate that the fractal dimensions of the different graphoglyptid ichnogenera are well constrained. The only obvious anomaly is that of a low fractal dimension for one specimen of Spirorhaphe, which probably was due to poor preservation of that particular specimen. The graphoglyptid fractal dimensions ranged from 1.277 to about 1.626, with each particular ichnogenus having a fractal dimension range of 0.097 to 0.147. 51 2.6 Conclusion The inferred behavioral significance of graphoglyptid burrows has been a debatable issue among ichnologists. A quantifiable means for comparing various kinds of trace fossils can be achieved by using a fractal geometric approach for analyzing trace fossils that represent different feeding behaviors. The results of the fractal analyses in this study show that pascichnial and fodinichnial behavior patterns display a high fractal dimension, whereas the graphoglyptid traces display a much lower fractal dimension. This observation indicates that graphoglyptid feeding patterns do not represent the maximum coverage of sediment, as would be expected for a deposit feeding strategy. The best-supported hypothesis for the graphoglyptid feeding patterns, therefore, is that they represent an agrichnial (farming) behavior. This study demonstrates that fractal analysis offers a useful methodology for ichnological interpretation. Fractal analysis can be used successfully in determining feeding behavior, and it also may be a helpful approach for determining ichnotaxobases of similar trace fossils. When examining the fractal characteristics of individual ichnogenera, each ichnogenus apparently has a narrow fractal dimensional range, although more analyses of graphoglyptids from different time periods and formations would help support this. Fractal analysis also is able to provide information on incomplete specimens. As long as the amount of missing information due to erosion is minimal, fractal analysis can be a powerful ichnological tool in the paleontologist's tool CH A PT ER 3 1 PITFALLS, TRAPS, AND WEBS IN ICHNOLOGY: TRACES AND TRACE FOSSILS OF AN UNDERSTUDIED BEHAVIORAL STRATEGY 3.1 Abstract The trapping of prey, where predators use external resources to help capture prey, is a specialized type of feeding behavior that is identified in the trace fossil record only rarely. Trapping traces that have been reported in the literature include spider webs, ant-lion burrows, scorpion pits, cerianthid tube anemone burrows, echiuran worm burrows, polychaete worm (Paraonis) burrows, and deep-sea graphoglyptids burrows. There is uncertainty, however, if all of these examples actually represent traps. Paraonis burrows, for example, have been represented as trapping traces, but there is a question if they actually represent this kind of behavioral strategy. Previous references and new field work indicate that Paraonis likely employs a selective deposit feeding strategy. In the fossil record, most of the known trapping traces are represented by spider webs, which are preserved in amber, and graphoglyptid burrows. Trace fossils that could represent trapping strategies may exhibit some basic morphological attributes, including (1) a 'Reprinted from Palacogcograpln. Palaeoclimatology, Palaeoecology, v. 375, James R. Lehane and A.A. Ekdale, Pitfalls, traps, and webs in ichnology: Traces and trace fossils of an understudied behavioral strategy, pp. 59-69, 2013, with permission from Elsevier. 53 conical depression composed of loose sediment; (2) an open pit; (3) a physical snare composed of a sticky substance; and/or (4) adequate spacing between the burrows, pits, or snare material without much overlapping. The interpretation that at least some graphoglyptids (e.g., Spirorhaphe) represent trapping was based on a trapping model for Pcircionis, but since Paraonis does not trap prey, the question arises whether graphoglyptids should be considered trapping at all. The variety of graphoglyptid morphologies supports the idea that graphoglyptids were not all doing the same thing. Previously, the ethological category of agrichnia has been applied to graphoglyptids and has been used to denote both trapping and farming behaviors, although the two behaviors display distinctly different feeding strategies. Some graphoglyptids may represent farming traces, while others may represent trapping traces, but it is unlikely that an individual burrow represented both behaviors. The new behavioral category ‘irretichnia' is proposed here to encompass trapping trace fossils, due to its unique behavioral significance and also to separate trapping from farming. 3.2 Introduction The trapping of prey is a highly specialized feeding behavior in the animal kingdom. Trapping involves the employment of external resources to help a predator capture prey. Traps can include sticky materials, pits to fall into, or any other activity where the predator does not search for and subdue the prey, but rather waits and ensnares the prey. Trapping does not include animals that use burrows or other structures as ambush points to attack prey. In this study, prey is considered as an animal (multicellular heterotroph) that is captured and consumed by a predator. 54 Among the most widely recognized examples of true trapping behavior by invertebrates are spiders that capture prey in intricately constructed webs. Another type of trap, exemplified by modem ant-lion burrows, has even made its way into the movies (as featured as the sandy ‘sarlaac' pit in the film, Return o f the Jedi). Even though some examples of predator traps, such as spider webs, are common in the modem environment, confirmed cases of predator traps in the fossil record are exceedingly rare. The purpose of this paper is to illustrate all of the cited examples of trapping in the animal kingdom, both modern and ancient, assess their identification in the modem and ichnological record, and analyze some known trace fossils that could represent trapping behavior. The designation of a new behavioral category "irretichnia" is introduced here to represent trapping traces in the trace fossil literature. 3.3 Modern trapping traces and their fossil equivalents 3.3.1 Ant-lion burrows Ant-lion larvae (Insecta, Myrmeleontidae) build traps by creating conical, pit-like burrows in loose sand. When a hapless ant stumbles into the burrow, the ant-lion proceeds to kick-up sand, preventing the ant from climbing out (Turner, 1915; Heinrich and Heinrich, 1984). The hapless ant then is eaten by the ant-lion. Several different species of ant-lions produce a range of pit morphologies (Fig. 3.1 A), some with V-shape walls and others with nearly vertical walls (Griffiths, 1980, 1986). Ant-lion burrows have not yet been identified in the fossil record. Ant-lions are known from the body fossil record since the Early Permian (Rasnitsyn and Quicke, 2002), but their feeding on ants probably has been a more recent adaptation, since ants 55 A B Figure 3.1. A, Cross-sectional view o f ant-lion burrows (based on discussion by Griffiths, 1980). Scale bar is ~1 cm. B, Cross-sectional view o f scorpion pits and burrows (modified from Harington, 1977). Scale bar is ~3 cm. are known only from the Albian (Heads et al., 2005). This o f course does not preclude the possibility that ancient ant-lions may have fed on other kinds o f insects in the same manner as they feed on ants today. An ant-lion burrow likely produces a V-shaped trace fossil that cuts across the layers of the sediment and is filled with unstratified sediment. There is one example of a burrow that has these characteristics from the Devonian (Morrissey et al., 2012), far earlier than the known occurrences o f ant-lion body fossils. 3.3.2 Scorpion burrows and pits Scorpions (Chelicerata, Scorpionidae) create multiple types o f burrows ranging from large, low-angle, branching burrows to vertical-walled pits, several o f which are constructed to act as traps (Williams, 1966; Shorthouse and Marples, 1980; Harington, 56 1977; Hembree et al., 2012). The longer scorpion burrows with gently sloping entrances usually serve as shelter from predators, but the openings also may serve as traps for passing prey (Hembree et al., 2012). The scorpion sits just inside the burrow openings to wait for the prey. The scorpion then attacks as the prey enters the burrow while trying to escape the high daytime surface temperatures (Shorthouse and Marples, 1980). Vertical scorpion burrows, which are constructed to be pitfall traps, are made by Cheloctonus jonesii (Pocock, 1892). Unlike the low-angle scorpion burrows, C.jonesii digs the pit and then returns later to retrieve the prey that has fallen in (Harington, 1977). The walls of the pitfall trap range from vertical to inwardly inclined, with the base of the trap being larger in diameter than the opening (Fig. 3.IB). Scorpion burrows are produced in a firmer substrate than the ant-lion burrows, including soils and firmer sands that can be compacted to stabilize the burrows. Terrestrial scorpion body fossils are known since the Early Mississippian (Kjellesvig-Waering, 1986). Most species of scorpions today create a burrow where the cross-section of the entrance has a flat bottom and crescent-shaped upper half, mirroring the cross-section of a scorpion body. This type of burrow is unique to scorpions, whereas most other burrowing animals create circular or oval burrow openings (Polis et al., 1986). Scorpion burrows are extremely rare in the fossil record, only being recorded in the Pleistocene of Sonora, Mexico (Phelps, 2002), but there probably are many that are unrecognized in the ichnofossil record (Hembree et al., 2012). Identification of many scorpion burrows in the fossil record should be straightforward if the observer knows what to look for, since they have a unique geometry (Phelps, 2002). 57 3.3.3 Spiderwebs Spiders (Chelicerata, Araneidae) produce sticky strands o f ‘silk' that many species, most commonly orb-weaver spiders, weave into a web for trapping prey (Kaston, 1964). The strength, structure, and shape of the webs are often taxon-specific, leading some researchers to infer that lineages of spider taxa might be determined using spider web morphology (Eberhard, 1990). Spiders have been producing silk since at least the Middle Devonian (Selden et al., 2008), and the silk produced in modern spiders is as strong as bulletproof clothing (Vollrath and Knight, 2001), thus rendering silk an effective material to be used in construction of a trap. In spite of the long history of spiders and the extreme strength of spider silk, spider webs are extremely rare in the fossil record. The problem is that even though spider silk is very strong, it also is very biodegradable and rarely preserved, except in amber, which is where all of the known fossil spider webs are found. Of the few findings of spider silk in the fossil record, the oldest is the occurrence of silk still attached to a spider from the Middle Devonian (Selden et al., 2008). The first reports of spider webs, or at least silk strands, occur in the Early Cretaceous (Zschokke, 2003; Jarzembowski et al., 2008; Brasier et al., 2009), followed by fossil silk found in the Middle Cretaceous, Eocene, and Miocene (Zschokke, 2004). Most of these examples are single threads, not full webs. There are a few reports of branching threads (Penney, 2008; Brasier et al., 2009) that might be construed as a partial trapping structure, and there are even fewer reports of fossilized spider webs with evidence of the web being used as a trap (Poinar and Poinar, 1999: figs. 70 and 71; Poinar and Buckley, 2012). 58 3.3.4 Cerianthid tube anemone burrows Cerianthid tube anemones (Cnidaria, Anthozoa, Cerianthidae) are stationary predators that live in short vertical burrows in intertidal sand flats, as seen in Cholla Bay, Sonora, Mexico (Fig. 3.2A). The burrowing anemone completely covers itself and waits for an unsuspecting victim to crawl across its camouflaged oral disk of poison-laden tentacles, where the victim is killed and ingested by the predator (Fig. 3.2B). Burrows of sea anemones have been identified in the trace fossil literature, often referred to the ichnogenera Conostichus and Bergaueria (Fig. 3.3). Even though at least some Conostichus and Bergaueria could be regarded as predatory trapping traces, these trace fossils are commonly referred to as anemone dwelling traces (domichnia) (Alpert, 1973; MacEachern and Pemberton, 1992; Mata et al., 2012). In fact, infaunal anemones have been identified as possible predators as early as the Cambrian, based on trilobite fragments found in the central portion of the trace fossil Dolopichnus gulosus (Alpert and Moore, 1975). The paleoethologic interpretation o f D. gulosus is that the predatory anemone used its burrow as a trap to catch, subdue, and ingest trilobite prey. 3.3.5 Echiuran worm burrows Even though the majority of marine organisms are denser than salt water (1,075 kg/m3 versus 1,026 kg/m3 for salt water), the ability of a marine predator to set up an effective pitfall trap for prey could be a problem due to some degree of buoyancy preventing the prey from sinking quickly (Alexander, 1990). One of the potential answers to counter prey buoyancy is a form of feeding done by the echiuran worm, Urechis caupo (Fisher and MacGinitie, 1928). This worm feeds by producing a mucus net across the top 59 Figure 3.2. Burrowing sea anemone at Cholla Bay, Sonora, Mexico. A, Sea anemone removed from the sediment. B, Sea anemone in its burrow and covered with sediment. Scale bar is 1 cm. Figure 3.3. Bergauerici isp. from Lower Cambrian Brigham Quartzite of Two Mile Canyon, Idaho (Specimen UUIC-1428 in University of Utah Ichnology Collection). A, Top view. B, Side view of the polished right hand portion of the cut surface. The arrows indicate the same location along the cut surface of the sample. Scale bar is 2 cm. 60 of one end of its U-shaped burrow, functioning in a similar way to a spider web (Ricketts and Calvin, 1968). The worm generates a suction current through the burrow by the use of rhythmic contractions, which causes water to be drawn across the net trapping plankton and other food particles (MacGinitie and MacGinitie, 1968). U. caupo then proceeds to eat the mucus net containing the trapped food all at once. The feeding methods of U. caupo are sometimes considered to be a type of filter feeding, even though an external net is employed. The mucus net combines some of the pertinent features of a scorpion pit and ant-lion burrow with a spider web into one structure, because the prey animal gets stuck in the net rather than falling passively into a pit. Burrows of echiuran worms have been identified in the trace fossil literature. Diplocraterionparallelum var. quadrum (Ekdale and Lewis, 1991) is a U-shaped burrow with a spreite from the Late Quaternary of New Zealand that very closely resembles modern echiuran burrows in the same area (Ekdale and Lewis, 1991). This trace fossil almost certainly is a fossil example of an echiuran trapping burrow. There is a possibility that some of the other U-shaped burrows in the trace fossil record (e.g., Diplocraterion and Arenicolites) also may have been created as trapping burrows by echiurans, or other similar organisms. 3.3.6 Paraonis burrows Paraonis fidgens (Levinsen, 1882) is an infaunal polychaete annelid worm that lives in sandy, intertidal zones along the coasts of North America and Europe today (Papentin, 1973; Risk and Tunnicliffe, 1978; Gaston et al., 1992). Previous reports in the literature have stated that Paraonis creates an open burrow system to facilitate a 61 specialized diet of benthic diatoms by using the burrows as a trap for diatoms. The Paraonis trapping model suggests that when the worm burrows through the sediment, it selects the diatoms by pushing the non-organic sediment grains aside, and at the same time it solidifies its burrow walls with mucus (Risk and Tunnicliffe, 1978). At a later point in time, live diatoms migrate through the sediment and become trapped in the mucus-lined burrows, allowing the worm to return through its burrow and feed on the trapped diatoms (Roder, 1971; Seilacher, 1977; Risk and Tunnicliffe, 1978). ROder's (1971) initial trapping model for Paraonis was based on the the geometrically consistent pattern of the burrows, the few burrow intersections, and the fact that the the majority of the Paraonis gut contents were benthic diatoms (along with smaller amounts of foraminifer, small crabs, soft-bodied metazoans, and green algae). Paraonis worms are found primarily in sandy substrates (96-99% sand) in littoral to sublittoral intertidal environments and are restricted to temperate latitudes (Gaston et al., 1992). The burrows of Paraonis can be found within oxic to anoxic sediments (Roder, 1971). In the oxic and dysoxic sediments, Paraonis worms create open burrow systems that are constructed as horizontal to semi-horizontal spirals, which emanate from the center and spiral outward (Fig. 3.4; Fig. 3.5A). The burrow path then continues in a horizontal plane spiraling outwards (Spirorhaphe-type) and will sometimes turn back upon itself following the initial curve of the spiral (Helminthoida-type). Within the sediment, the burrow system will often consist of multiple spiral tunnels tiered on top of each other. These spirals can be sufficiently dense as to completely bioturbated the sediment (-100 burrows/1,000 cm3). In the deeper, anoxic sediment, Paraonis creates semi-vertical branching burrows (Roder, 1971). 62 Figure 3.4. View of tiered structure o f burrows o f Paraonis fulgens (Levinsen, 1882), a polychaete annelid (modified from Roder, 1971: fig. 11). The 10 cm depth illustrated here indicates where the spiral burrows are located within the oxic and dysoxic sediments. Below 10 cm is the anoxic sediment. The speckled pattern at the top o f the box represents the topmost centimeter, the only level where living diatoms are found. Some questions arise regarding the Paraonis trapping model. Roder (1971) stated that the spirals were built quickly and were short-lived (>76% o f the burrows were destroyed within five days), although the Paraonis worms were reported to have traveled along the burrows multiple times within those five days. There were no diatoms discovered within the mucus lining o f the burrows, and no vertical migration o f living diatoms was witnessed, although there was an assumption that the light needed for the investigation stalled any migration (Roder, 1971). There also is evidence that the Paraonis worm possibly was not feeding on a specialized diet o f diatoms after all. The 63 Figure 3.5. Paraonisfulgens burrows at Goose Point, Willapa Bay, Pacific County, Washington State. All of the photos were taken at SL3. A, Pair of typical Spirorhaphe-type Paraonis burrows. B, Sediment cross section showing Paraonis burrows and cross­cutting vertical worm burrows with oxidized burrow margins. C, Subhorizontal cross­section of the sediment illustrating multiple burrows covering different depths of the sediment from one to six centimeters in depth. Scale bars are 3 cm. food discovered in the guts of some Paraonis worms was more representative of a deposit feeding organism containing mostly empty diatom frustules, dinoflagellates, and detritus (Gaston et al., 1992). To explain the reason that the worms analyzed by Roder (1971) had such a high concentration of diatoms in them, there was a suggestion that the gut contents of the worm simple reflected the sediment composition and not necessarily the "feeding method" of the worm. Paraonis burrows are found in several localities in modern environments, including the North Sea (Roder, 1971), Gulf of Saint Lawrence (Brunei et al., 1998), Bay of Fundy (Risk and Tunnicliffe, 1978), Gulf of Mexico (Gaston et al., 1992), and Willapa 64 Bay in Washington State, USA (Gingras et al., 1999). Contrary to suggestions from previous workers (Risk and Tunnicliffe, 1978), there appear to be few factors that favor preservation of an open burrow system in a dynamic intertidal environment, so there is not an expectation that these burrows would be prevalent in the fossil record. There is one report of a Paraonis-like burrow in the fossil record from Permian tidal flat deposits in New Mexico (Minter et al., 2006). The spacing of the spiraling tunnels in the Permian fossil are much closer than has been observed in modern examples, but the general morphology is similar. In order to determine which of the feeding habits was more likely (trapping or deposit feeding) another investigation of the burrows was needed. Therefore, for this study, burrows of Paraonis were observed at Willapa Bay in Washington State (Fig. 3.6). 3.3.6.1 Study area. Willapa Bay is a mesotidal (2-3 m tidal range) estuary that is separated from the Pacific Ocean by the North Beach Peninsula, a 27 km-long spit in the state of Washington. Paraonis burrows were observed at Goose Point along the eastern shore of the bay and were restricted mostly to the middle to lower intertidal deposits along Goose Point (Figs. 3.6-3.8). There were no upper intertidal deposits in this area owing to the edge of the tidal flat ending in a sea wall along the coast. The middle intertidal zone was composed of compacted sand with the upper 1 cm composed of oxygenated sand and the middle 2-10 cm composed of dysoxic sand. The Paraonis burrows did not exhibit any oxidized burrow margins but cross-cutting vertical worm burrows did show some evidence of oxidization (Fig. 3.5B). The lower intertidal zone was composed of loose sand and was similar to the middle intertidal zone in that the upper 1 cm was oxygenated and the middle 2-10 cm was dysoxic (Fig. 3.9). In both 65 Figure 3.6. Location of modem Paraonis burrows at Goose Point, Willapa Bay, Pacific County, Washington State. The star indicates the sampling locations (SL) featured in Figure 3.7. zones, the sand was composed o f -90% fine sand with negligible clay content, and the composition is mainly quartz in both locations. Burrows were located by digging in rippled sand, free of eelgrass (Zostera marina), and with a lower concentration of open burrows on the surface (which were produced by decapod crustaceans and other polychaete worms) in the lower and middle intertidal zones. The burrows could be found by inserting a shovel into the sediment and pulling out clumps of sand. The sand clumps were broken by hand along sedimentary layering, horizontal to the surface. The Paraonis burrows were sufficiently distinct from other burrows to be identified easily (Gingras et al., 1999). Some Paraonis burrows were not perfectly horizontal, but instead they appeared to be following subhorizontal laminae 66 Figure 3.7. Field map of sampling locations (SL) for Paraonis burrows at Goose Point, Willapa Bay, Pacific County, Washington State. Stars indicate sites where Paraonis burrows were identified. The dotted line identifies the approximate divide between different tidal zones. within the rippled sand. Burrows also could be identified as a line of small holes along the broken edge of the sand clump (Fig. 3.9). 3.3.6.2 Observations. Paraonis burrows were found from within the top centimeter of the sediment to 10 cm below the surface. In the lower intertidal zone, the burrows frequently were found isolated and very friable, meaning the burrows fell apart easily if the sediment was disturbed. The friability of the sediment made identification of the deeper individual burrows more difficult, so most of the identified burrows occurred within the top 4 cm (Table 3.1). One to two burrows were found within each clump at Figure 3.8. Photo of the intertidal flats at Goose Point, Willapa Bay, Pacific County, Washington State. Photo was taken at low tide at site SL3, facing west towards the current shoreline. Shovel for scale. various depths. In the middle intertidal zone, the burrows were identified in one location free of eelgrass in harder packed sand where the cohesion of the sand grains prevented slight disturbances from destroying the burrows. Burrows in the middle intertidal zone were spaced closer together than the lower intertidal zone. The burrows were so abundant that they seemed to pervade the sediment, with burrows stacked within centimeters on top of each other. Due to the shear abundance of Paraonis burrows in the middle intertidal zone, they were not counted at different levels within the sediment, but every cross section for the 17 different holes analyzed had approximately 5-10 Paraonis burrows per 100 cm2 on each horizontal surface. 68 Figure 3.9. Sediment cross-section containing Paraonis fulgens burrows at Goose Point, Willapa Bay, Pacific County, Washington. The cross-section is from the lower intertidal zone taken at SL2. Arrow highlights the burrows as seen in cross-section. Dashed line indicates the transition from oxic to dysoxic sediment. Scale bar is 1 cm. Throughout the intertidal zone, the burrows were found most often as portions of circular burrow networks extending outward into meandering arcs. Both types of Paraonis burrows (Spirorhaphe-type and Helminthoida-type) were identified in the field. Close inspection of the burrow walls with a hand-lens did not identify any supporting mucus or other material. Even though there was no mucus was visible microscopically, there is a possibility that small amounts of mucus may be lining the walls. Individual 69 Table 3.1. Number of Paraonis burrows identified within the lower intertidal zone at Goose Point, Willapa Bay, Pacific County, Washington State at -4 0 studied locations. Depth below surface (cm) Number of Paraonis burrows 0-1 2 1-2 8 2-3 9 3-4 5 4-5 1 5-6 0 6-7 0 7-8 0 8-9 0 9-10 1 Unsure 2 burrows were -0.5 mm wide with -2 mm spacing between each whorl. Burrow networks ranged in width from 1 cm to several centimeters. Paraonis worms were found in the sediment, although there were no worms found within the Paraonis burrows or observed in the process of making any of the burrows. 3.3.6.3 Discussion - Testing the Paraonis trapping model. The ethologic model of Paraonis burrows as diatom traps is questionable. Diatoms have been shown to burrow in the sediment, although they typically only burrow down a few millimeters, up to a maximum depth of 1.4 cm. Diatoms also tend to live in sediment with higher silt and clay content because of the water that is retained (Hay et al., 1993; Aleem, 1950). The deepest Paraonis burrow (10 cm) is over seven times deeper than the depth of the deepest burrowing diatom. This indicates that although the shallowest burrows of Paraonis could be diatom traps, any burrow deeper than 1.4 cm would not function as such. If Paraonis burrows functioned as traps, then the Paraonis worms must be using them to trap some type of prey other than diatoms. 70 Another problem with the trap model is that the burrows would need to withstand some disturbance while the Paraonis worm re-enters the burrow in order to feed. Previous observations indicate that most of the burrows are destroyed within five days (ROder, 1971). Personal field observations indicate that there is not enough mucus, if any, to support the burrows as open tunnels, so that the burrows in looser sediment collapse with minimal disturbance. The sediment is readily disturbed by the shifting tides, which move the sand within the tidal flat twice a day. Even though it has been hypothesized that Paraonis revisited the burrows to feed on the diatoms which got stuck in the mucus-lined walls, there has been no observed proof of any diatoms trapped in mucus (Roder, 1971). The morphology of the burrows also is inconsistent with a trapping model. For the worm burrows to act as a trap, the individual nets would need to be semi-isolated. Stacking the nets on top of each other could render the central nets useless, since passing organisms would get stuck on the outer nets. The Paraonis burrows at Willapa Bay are stacked upon each other in layers and packed so tightly together in some spots as to completely permeate the sediment (Fig. 3.3B). All of this evidence indicates that the Paraonis worm was not trapping its prey, as has been suggested, but was likely a selective deposit feeding worm. The open burrow system was produced as a side effect of the worm moving through the grains selecting out the individual food items that it wanted, while pushing the inorganic sediment aside. The mucus on the body of the worm helped it slide through the sediment causing the burrows to remain open temporarily, even though the worm would likely not need to travel back through the open burrows. 71 3.4 Possible ancient traps The few trace fossils representing undoubted examples of traps identified in the fossil record include reports of fossilized spider webs in amber (Penney, 2008; Brasier et al., 2009), a Devonian funnel-shaped pit that has been cited as a possible analogue to a modern ant-lion burrow (Morrissey et al., 2012), and Pleistocene scorpion burrows (Phelps, 2002). There is a possibility that many other trace fossils described in the literature, whose overall morphology resembles that of modern trapping pits and burrows, may have been misinterpreted. There are several common biogenic sedimentary structures, including some in the marine realm, which could be reinterpreted as trapping structures. For a trace fossil to be interpreted as a trap, the trace fossil would likely possess several key features, including one or more of the following: (1) a conical depression composed of loose sediment; (2) an open pit; (3) a physical snare composed of a sticky substance; and/or (4) adequate spacing between the burrows, pits, or snare material without any overlapping. Snare material would not be preserved in most cases, except via preservation in amber, but there is a possibility that some structures could indicate the former presence of snare material, like the mucus used to compose the tube walls of many annelid worms (Ekdale et al., 1984a). Spacing is needed in a spider web to provide the largest net possible while using the least amount of material. Spacing is needed in burrows and pits to provide a ‘landmine' approach to the field where close spacing would work counter to this by preventing more prey from entering the ‘minefield'. While the previously mentioned criteria apply to terrestrial traps, they also may be extended to the 72 marine realm. Several existing marine ichnotaxa fit the criterion of a conical depression filled with loose sediment. 3.4.1 Simple pits and burrows The simplest type of trapping trace is a pit of the kind created by ant4ion larvae and certain scorpions. These can be formed in loose sand with a conical shape (ant-lion like) or in a firmer substrate with vertical walls (scorpion like). Even though such pits commonly are found in a terrestrial environment there is a possibility that pits in the marine environment also could represent trapping traces. Conical marine trace fossils, which would mimic the ant-lion method of trapping, include the following (Table 3.2): Monocrater ion, a ‘funnel-shaped' trace fossil (Goodwin and Anderson, 1974); Conichnus (Fig. 3.10), a conical ichnofossil that contains cone-in-cone chevron laminations that do not widen upwards (Myannil, 1966; Weissbrod and Barthel, 1998; Buck and Goldring, 2003); Conostichus, a cone-in-cone structure with interbedded sandy and muddy layers that contained concentrated sand around the outer walls of the funnel (Chamberlain, 1971; Pfefferkorn, 1971); Cormrfatichnus, a sub-vertical conical burrow with massive infilling that is interpreted as a subaqueous open burrow (Carroll and Trewin, 1995); and Altichnus, a funnel-shaped tube that is always oriented perpendicular to the surface (Gaillard and Racheboeuf, 2006). Other possible trapping structures include un-named escape traces (Fugichnia), ‘funnel-shaped' structures (Weissbrod and Barthel, 1998; Magyar et al., 2006; Jamer et al., 2011), and ‘cone-in-cone' collapse structures (Buck and Goldring, 2003), where the inward collapse of the sediment was the result of depression in the underlying sediment. 73 Table 3.2. List of possible trapping traces and structural features previously mentioned in the trace fossil literature. Trace fossil or Structure Description Alticlmus Bergaueria Cone-in-cone structures Conichnus Conostichus Cornulatichnus Diplocraterion parallelum var. qua drum Dolopichnus gulosus Fugichnia structures Funnel-shaped structures Monocraterion a funnel-shaped tube that is always oriented perpendicular to the surface Conical trace fossil inward collapse of the sediment was the result of depression in the underlying sediment conical ichnofossil that contains cone-in-cone chevron laminations that do not widen upwards cone-in-cone structure with interbedded sandy and muddy layers that contained concentrated sand around the outer walls of the funnel sub-vertical conical burrow with massive infilling that is interpreted as a subaqueous open burrow U-shaped burrow with a spreite Conical trace fossil with trilobite fragments Funnel-shaped structures interpreted as the result of a buried organism trying to escape Funnel-shaped trace fossils with unknown behavioral significance ‘funnel-shaped' trace fossil 3.4.2 Graphoglvptid burrows Graphoglyptids trace fossils have been interpreted as representing traps and/or farming burrows (agrichnia) by several workers (e.g., ROder, 1971; Seilacher, 1977; Miller, 1991b; Uchman, 2003; Minter et al., 2006). Graphoglyptid burrows are geometrically complex, predepositional, open burrow systems, commonly preserved in positive hyporelief on the soles of deep-sea turbidite beds (Fig. 3.11). The taxonomic affinities of the producers of graphoglyptids are unknown, but they most likely represent the work of some type of worm or arthropod (Garlick and Miller, 1993). The geometric shapes of graphoglyptids range from meanders (Cosmorhaphe) to spirals (Spirorhaphe) to intricate networks (Paleodictyon). The width of individual graphoglyptid burrow tunnels ranges Figure 3.10. Conichnus conicus (Myannil, 1966) from Middle Ordovician Yiyhvits Horizon of Estonia (Specimen UUIC-1148 in University of Utah Ichnology Collection). Scale bar is 1 cm. from approximately 0.1 cm to greater than 5 cm (Uchman, 2003). Modem graphoglyptids are found within the top few centimeters of deep-sea sediment (Ekdale, 1980), which is characterized by three zones: the Mixed Layer, the Transition Layer, and the Historical Layer (Berger et al., 1979; Ekdale et al., 1984b). Graphoglyptid burrows are formed within the uppermost couple of centimeters of the Mixed Layer, which extends down from the sediment-water interface to depths of 3 to 10 cm. This uppermost zone of the deep-sea sediment typically is completely homogenized by very active burrowers, so in a continuously accreting pelagic substrate, graphoglyptid burrows are not preserved (Ekdale et al., 1984b). 75 Figure 3.11. Photos of selected graphoglyptids. A. Cosmorhaphe from the Lower Eocene Guipuzcoan Flysch of Zumaia, Spain. B. Helicolithus from the Lower Eocene Guipuzcoan Flysch of Zumaia, Spain. C. Helminthorhaphe from the Upper Cretaceous of Tanzania (Specimen UUIC-1911 in University of Utah Ichnology Collection) D. Megagrapton from the Lower Eocene Guipuzcoan Flysch of Zumaia, Spain. E. Paleodictyon from the Jurassic of Calabria, Italy (Specimen UUIC-1164 in University of Utah Ichnology Collection). F. Spirorhaphe from the Upper Cretaceous of Tanzania (Specimen UUIC-1902 in University of Utah Ichnology Collection). Scale bars are 4 cm. Unlike the other examples of possible trapping, graphoglyptid burrows have been recognized primarily from the fossil record, with modem examples only being discovered within the last 35 years (Ekdale, 1980). Graphoglyptid trace fossils are known since the Cambrian, but the producers of graphoglyptid burrows have not yet been identified in fossil or recent occurrences of the burrows (Ekdale, 1980; Rona et al., 2009). The graphoglyptid trapping model is based on the interpretations of Roder (1971) for Paraonis burrows. The model was extended by Roder (1971) and Seilacher (1977) to interpret graphoglyptids, who cited the following similarities between Paraonis and graphoglyptids: (1) they are both open burrow systems; (2) the sharp outlines of fossil graphoglyptids indicated that the walls were likely reinforced by a "stronger than usual 76 mucus film" ; (3) branching has been observed in both graphoglyptids and Paraonis burrows, which would be counterintuitive for a deposit feeding strategy; (4) careful distance is maintained between the burrows in both groups; and (5) the is no backfill in either group of burrows, which would indicate deposit feeding and would prevent turning around and retreading through the burrows. Since that time, the trapping model has been expanded to include all graphoglyptids, no matter the shapes (e.g., Miller, 1986, 1991b; de Gibert et al., 1999; Uchman, 2003). 3.4.2.1 Questioning the graphoglyptid trap model. For the trapping model to explain the behavior represented by graphoglyptid burrows, the graphoglyptid trace producers must be feeding on organisms that migrate through the sediment. The prey organisms must be small enough to get trapped in an open graphoglyptid burrow (<0.1 cm). Unlike the trapping model for Paraonis that is based on diatoms, photosynthetic protists like diatoms are immediately ruled out as potential graphoglyptid prey, since the deep-sea graphoglyptids occur far below the photic zone. Some of the potential prey organisms that could be caught in a trap are the ones that are mixing the sediment in the Mixed Layer. These organisms are the meiofauna (between 60 and 300 |iin), which includes foraminifera, nematodes, harpacticoid copepods, polychaetes, ostracods, and crustacean nauplii (Wolff, 1977). Paraonis burrows have b