The Pennsylvania State University

The Graduate School

Department of Geosciences

MOLECULAR AND ISOTOPIC INDICATORS OF CANOPY CLOSURE

IN ANCIENT FORESTS AND THE EFFECTS OF

ENVIRONMENTAL GRADIENTS ON LEAF ALKANE EXPRESSION

A Dissertation in

Geosciences and Biogeochemistry

by

Heather V. Graham

© 2014 Heather V. Graham

Submitted in Partial Fulfillment of the Requirements for the degree of

Doctor of Philosophy

August 2014 The dissertation of Heather V. Graham was reviewed and approved* by the following:

Katherine H. Freeman Professor of Geosciences Dissertation Advisor Co-Chair of Committee

Lee R. Kump Professor of Geosciences Co-Chair of Committee

Peter D. Wilf Professor of Geosciences

Mark E. Patzkowsky Associate Professor of Geosciences

Erica A.H. Smithwick Associate Professor Ecology Outside Committee Member

Scott L. Wing Curator of Paleobotany, Smithsonian Institution Special Member

Chris J. Marone Professor of Geosciences Associate for Graduate Programs and Research in Geosciences

*Signatures are on file in the Graduate School

ii ABSTRACT

Dense forests with closed canopies are of enormous climatic and ecological significance. Three-dimensional forest structure affects surface albedo, atmospheric circulation, hydrologic cycling, soil stability, and the carbon cycle, both on the local and global scale. Increasingly, dense canopy forests are recognized for their role in plant and animal evolution and provide a great variety of microhabitats into which much of the diversity of animal and plant life has specialized. The geologic history of three dimensional forest structure is poorly known, however, because fossil assemblages seldom reflect tree spacing, size, or branching and adaptations specific to closed-canopy forests – such as large, fleshy fruits or branchless boles – rarely preserve. The best preserved and most abundant plant fossils are leaves and leaf fragments, however leaf morphology alone does not indicate canopy conditions. This study focuses on leaf biochemical signatures that have potential to record light environment, preserve in geologic archives, and ultimately serve as proxies of canopy density.

Dense, closed-canopy forests are defined by strong vertical light and humidity gradients, and a pronounced decrease in photosynthetic rate in the understory. This rate

13 difference results in an enrichment in the carbon isotope composition of leaves (δ Cleaf) at the top of the canopy relative to the lower levels. Our study found that leaves from a

13 deeply-shaded closed-canopy forest in Panama had a ~10‰ vertical range in δ Cleaf values, vertically, while leaves from a seasonal canopied forest in Maryland expressed

13 only a ~6‰ range. A Monte Carlo model that resampled δ Cleaf values indicated that

iii canopy closure could be identified by isotopic range from a relatively small (~50) number

13 of leaves. Based on this result, we used δ Cleaf range as a diagnostic feature of canopy

13 coverage and measured δ Cleaf values of fossil leaves to identify canopy density characteristics of three ancient forests. Our results confirm the earliest closed-canopy

Neotropical forest in a Paleocene fossil assemblage and identified open canopy features in a Cretaceous fossil assemblage and a forest edge in another Paleocene assemblage.

Identifying canopy closure in fossil leaves by analogy by isotopic leaf features observed in modern plants assumes that patterns of photosynthetic carbon isotope fractionation (∆leaf) are the same in ancient plants and their extant relatives. ∆leaf values calculated for modern leaves had a wide range of values from canopy to understory. ∆leaf values calculated from fossil leaf data indicated that the majority of leaves in the assemblages were from the upper canopy. A similar analysis of n-alkanes from modern

13 found that modern leaves express a wide range of δ Clipid values, in correlation with

13 humidity, canopy height, and irradiance. δ Clipid values from fossil leaves had smaller ranges than modern relatives but a mixing model that used the relative abundance of leaves

13 in the assemblage and the fossil δ Clipid data closely reproduced the chain-length

13 distribution and δ Clipid of alkanes from bulks sediments from the fossil assemblage.

Calculated fractionation during lipid synthesis (εlipid) in fossil leaves was less negative than modern leaf values and may reflect drier climates or the bias in fossil leaf assemblages toward upper canopy leaves.

iv The amount of n-alkanes made by a leaf is greater in the understory than it is in the

13 canopy. Understory leaves also have more depleted δ Clipid values. Alkanes extracted

13 from fossil leaves in the closed-canopy assemblage also exhibited depleted δ Clipid values in leaves with higher alkane content. An analysis of the published literature found leaf higher angiosperms produce greater concentrations of leaf alkanes. The systematic increase in alkane production as well as the higher alkane amounts in understory leaves may be related to the enhanced fungal resistance proffered by alkanes that would be necessary in humid understories. Increased alkane production may then also be related to shade tolerance and early adaptation by higher angiosperms to closed-canopy conditions.

v TABLE OF CONTENTS

LIST OF FIGURES…………………………………………………………………….. x

LIST OF TABLES……………………………………………………………………. xiv

ACKNOWLEDGEMENTS………………………………………………...………...xvii

EPIGRAPH……………….………………………………………………...……….....xix

Chapter 1. Introduction...... 1

1.1. Introduction...... 1 1.2. Climatic feedback in closed-canopy forests ...... 2 1.3. Reconstruction of ancient canopied forests from fossils...... 2 1.4. The origin and evolution of the closed-canopy forest in the Neotropics ...... 3 1.5. Biochemical features of canopy closure: The canopy effect...... 4 1.6. Isotopic reconstruction of ancient ecosystems...... 6 1.7. The advantages of plant-specific biomarkers in ecosystem reconstruction ...... 7 1.8. Ecophysiology of leaf alkane production...... 8 1.9. Factors affecting preservation of biomarkers ...... 9 1.10. Factors affecting organic matter preservation in leaf fossils...... 10 1.11. Isotopic notation and calculation of isotope fractionation...... 11 1.12. Study objectives and chapter outlines ...... 13 1.13. Publications arising from this dissertation...... 15 1.14. References...... 17

Chapter 2. Isotopic characteristics of canopies in simulated leaf assemblages ...... 27

2.1. Abstract...... 27 2.2. Introduction...... 28 2.3. Methods ...... 33 2.3.1. Collection sites ...... 33 2.3.2. Sampling methods and environmental measurements...... 34 2.3.3. Isotopic analysis methods...... 36 2.3.4. Flux and biomass...... 38 2.3.5. Bootstrap analysis...... 39 2.4. Results ...... 41 2.4.1. Light and height measurements...... 41 2.4.2. Leaf carbon isotope data ...... 41 2.5. Discussion...... 42 2.5.1. Light properties and the δ13C canopy effect at San Lorenzo ...... 42 2.5.2. Canopy isotope gradients in biomass and litter flux ...... 44

vi 2.5.3. Resampling model outcomes ...... 45 2.5.4. From leaves to soil organic carbon ...... 50 2.6. Conclusions ...... 51 2.7. Figures ...... 53 2.8. References ...... 61

Chapter 3. Biogeochemical confirmation of the first closed-canopy Neotropical forest in the fossil record: A novel application of isotopic methods ...... 70

3.1. Abstract ...... 70 3.2. Introduction ...... 71 3.3. Methods ...... 76 3.3.1. Collection sites ...... 76 3.3.2. Sampling and analytical methods ...... 79 3.4. Results ...... 81 3.5. Discussion ...... 83 3.5.1. Guaduas (Maastrichtian) ...... 84 3.5.2. Cerrejón Site 0315 ...... 85 3.5.3. Cerrejón Site 0318 ...... 86 3.6. Conclusions ...... 88 3.7. Figures ...... 90 3.8. References ...... 96

Chapter 4. Plant waxes and fossil leaves from Cretaceous and Paleocene Neotropics: a Comparison of Modern and Ancient Fractionation Factors ...... 102

4.1 Abstract ...... 102 4.2 Introduction ...... 103 4.3. Methods ...... 105 4.3.1. Collection Sites ...... 105 4.3.2. Sampling methods ...... 108 4.3.3. Organic carbon analyses ...... 110 4.3.4. n-Alkane analyses ...... 111 4.4. Results ...... 113 4.4.1 Carbon isotope composition of fossils and leaves ...... 113 13 4.4.2. δ Clipid for Fossils and Leaves ...... 114 4.5. Discussion ...... 115 4.5.1. Photosynthetic fractionation in modern and ancient leaves ...... 115 4.5.2. Lipid biosynthetic fractionation in modern and ancient leaves ...... 119 4.6. Conclusions ...... 122 4.7. Figures ...... 124 4.8. References ...... 134

vii Chapter 5. Simulated and observed alkane distribution and isotope composition in sediments using biogeochemical analysis of fossil leaves ...... 140

5.1 Abstract ...... 140 5.2 Introduction ...... 141 5.3. Methods ...... 143 5.3.1. Collection Sites ...... 143 5.3.2. Fossil organic material and bulk rock sampling and preparation ...... 144 5.3.3. Organic carbon analyses ...... 146 5.3.4. n-Alkane analyses ...... 147 5.3.5. n-Alkane mixing model ...... 149 5.4. Results ...... 150 5.4.1 n-Alkane yields ...... 150 13 5.4.2 δ Clipid for fossils and sediments ...... 151 5.5. Discussion ...... 151 5.5.1. n-Alkane yield from fossil leaves ...... 152 5.5.2. Alkane chain-length distribution patterns in fossil leaves ...... 152 5.5.3. n-Alkanes from sediments integrate n-alkane patterns from leaves ...... 153 5.6. Conclusion ...... 156 5.7. Figures ...... 157 5.8. References ...... 162

Chapter 6. Shade tolerance, evolution and the ecophysiology of alkane production in angiosperms ...... 165

6.1. Abstract ...... 165 6.2. Introduction ...... 166 6.2.1. Ecophysiology and alkane production ...... 168 6.3. Methods ...... 170 6.3.1. Vertical transect leaf sampling ...... 170 6.3.2. Alkane extraction and preparative chromatography ...... 171 6.3.3 Alkane identification and quantification ...... 172 6.3.4. Published data selection ...... 173 6.4. Results ...... 174 6.4.1. San Lorenzo, Panama ...... 174 6.4.2. Published data ...... 175 6.5. Discussion ...... 176 6.5.1. Closed canopies and the early angiosperms ...... 176 6.5.2. The function of alkanes on leaf surfaces ...... 177 6.5.3. Alkane concentrations related to other defense compounds in leaves ... 179 6.5.4. Phylogenetics and metabolite production ...... 182 6.6 Conclusion ...... 183 6.7 Figures ...... 185 6.8. References ...... 193

viii Chapter 7. Research summary ...... 201

7.1. Chapter summaries ...... 201 7.2. Future directions ...... 204 7.3. References ...... 206

Appendix A. Evapotranspiration, humidity and lipid synthesis ...... 207 A.1. Introduction ...... 207 A.2. Methods ...... 209 A.2.1. n-Alkane isolation ...... 209 A.2.2. Compound specific D/H ratio analysis ...... 209 A.3. Results ...... 211 A.4. Discussion ...... 212 A.5. Figures ...... 213 A.6. References ...... 215

Appendix B. Nitrogen economics and δ15N response to irradiance ...... 217 B.1. Introduction ...... 217 B.2. Methods ...... 218 B.2.1. Sample selection and preparation ...... 218 B.2.2. Isotopic Analysis ...... 218 B.3. Results ...... 219 B.4. Discussion ...... 219 B.5. Future work and implications ...... 221 B.6. Figures ...... 222 B.7. References ...... 225

Appendix C. Molecular and isotopic investigation of the Mar 2X core ...... 227 C.1. Introduction ...... 227 C.1.1. Canopy structure across the PETM ...... 228 C.1.2. Taxonomic shifts and their influence on the CIE ...... 229 C.2. Samples and methods ...... 229 C.2.1. Mar 2X core ...... 230 C.2.2. Preparation of rock from core samples ...... 230 C.2.3. Alkane extraction and chromatographic preparation ...... 231 C.2.4. n-Alkane identification, quantification and isotope analysis ...... 232 C.3. Results ...... 233 C.4. Discussion ...... 235 C.5. Figures ...... 236

Appendix D. Data Tables ...... 240

ix LIST OF FIGURES

Figure 2.1 ...... 54 Map of collection sites

Figure 2.2 ...... 55 Light measurements and foliar δ13C for both tropical and temperate canopy leaf sampling sites

Figure 2.3 ...... 56 13 Seasonal canopy standing biomass, annual leaf flux, and δ Cleaf data for 5m vertical bins for tropical and temperate forests

Figure 2.4 ...... 57 13 δ Cair and ∆leaf for each sampling site from the Panamá site as related to height within the canopy, % light availability, and relative humidity

Figure 2.5 ...... 58 Relative humidity as related to height within the Panamá site canopy and % light availability

Figure 2.6 ...... 59 Absolute isotopic ranges (∆y) returned by the resampling model

Figure 2.7 ...... 60 13 Foliar δ Cy-mean ranges returned by model for both canopy types

Figure 2.8 ...... 61 13 Predicted δ Clitter returned by the resampling model and four litterfall collection sites and six soil samples from a transect of Panamanian forest

Figure 3.1 ...... 91 Modern leaf collection site in Panamá site and the Guaduas and Cerrejón fossil collection sites in two locations in Colombia

Figure 3.2 ...... 92 Stratigraphic column of Cerrejón Formation from Wing et al., (2009)

Figure 3.3 ...... 93 13 Measured δ Cleaf range for all modern and fossil leaves

Figure 3.4 ...... 94 13 Histograms of measured δ Cleaf range for fossil leaves showing the 13 frequency of δ Cleaf values expressed in fossil leaves

x Figure 3.5 ...... 95 13 Absolute δ Cleaf range for each fossil leaf site and the predicted isotopic range for closed canopy litter based on sample size predicted in Chapter 2

Figure 3.6 ...... 96 Calculated ∆leaf values for modern leaves from three binned strata of forest and ∆leaf values for fossil leaves from three study localities

Figure 4.1 ...... 125 Modern leaf collection site in Panamá site and the Guaduas and Cerrejón fossil collection sites in two locations in Colombia

Figure 4.2 ...... 126 Stratigraphic column of Cerrejón Formation from Wing et al., (2009)

Figure 4.3 ...... 127 Mean % Total organic carbon for all leaves from assigned morphotypes at all three fossil localities

Figure 4.4 ...... 128 13 δ Clipid values for all alkane homologues extracted from fossil leaves. Data arranged by families for all morphotypes.

Figure 4.5 ...... 129 13 δ Clipid values for all alkane homologues extracted from fossil leaves. Data arranged by locality.

Figure 4.6 ...... 130 13 Measured δ C25-35 values for alkanes extracted from paired modern and fossil leaves

Figure 4.7 ...... 131 Calculated ∆leaf values for modern leaves from three binned strata of the forest and ∆leaf values

Figure 4.8 ...... 132 Calculated εlipid values for alkanes extracted from fossil and modern leaves. Modern data binned by understory or canopy habit.

Figure 4.9 ...... 133 Calculated ∆leaf values for modern leaves from three binned strata of the forest and ∆leaf values for fossil leaves arranged by morphotype

Figure 4.10 ...... 134 Total organic carbon yield for fossil leaves arranged by morphotype

xi Figure 5.1 ...... 158 Modern leaf collection site in Panamá site and the Guaduas and Cerrejón fossil collection sites in two locations in Colombia

Figure 5.2 ...... 159 Stratigraphic column of Cerrejón Formation from Wing et al., (2009)

Figure 5.3 ...... 160 Average chain length for alkanes extracted from Cerrejón Formation leaves paired with similar data from modern collections.

Figure 5.4 ...... 161 13 δ Clipid values for all alkane homologues extracted from fossil leaves. Data arranged by families for all morphotypes.

Figure 5.5 ...... 162 13 δ Clipid values, ACL, and alkane abundance for Cerrejón Formation 13 sediments compared with modeled δ Clipid values, ACL, and alkane abundance based on collected lipid data and published leaf census.

Figure 6.1 ...... 186 Map of collection site

Figure 6.2a ...... 187 Abundance of extracted alkanes in dry leaf material from Aspidosperma cruenta (Gentianales)

Figure 6.2b ...... 188 Abundance of extracted alkanes in dry leaf material from Brosimum utile (Rosales)

Figure 6.2c ...... 189 Abundance of extracted alkanes in dry leaf material from Calophyllum longifolia (Malpighiales)

Figure 6.2d ...... 190 Abundance of extracted alkanes in dry leaf material from Manilkara spp. (Ericales)

Figure 6.2e ...... 191 Abundance of extracted alkanes in dry leaf material from Virola sebifera (Magnoliales)

Figure 6.3 ...... 192 Average chain length of extracted alkanes in leaves from a compilation of published data, arranged by broad taxonomic association

xii

Figure 6.4 ...... 193 Abundance of extracted alkanes in leaves from a compilation of published data, arranged by broad taxonomic association with APG III phylogenetic tree

Figure A.1 ...... 214 D/H values for alkane n-C31 expressed as a function of light availability

Figure A.2 ...... 215 δ18O values for α-cellulose varies versus relative humidity

Figure B.3 ...... 223 Total nitrogen in leaf material plotted versus light availibility

Figure B.4 ...... 224 δ15N of leaves trends vertically with height from positive to negative

Figure B.5 ...... 225 δ15N values arranged by genera and canopy position

Figure C.6 ...... 237 %TOC from Mar 2X core samples through PETM interval

Figure C.7 ...... 238 Alkane concentrations in Mar 2X core normalized organic carbon

Figure C.8 ...... 239 Stratigraphic column for Mar 2x (from Jaramillo et al., 2010) with ACL of 13 extracted alkanes and δ CTOC values for each sampling site

Figure C.9 ...... 240 13 Weighted mean average δ C25-35 values for alkanes from the Mar 2X

xiii

LIST OF TABLES

Table D.1 ...... 241 Environmental and isotopic data from vertical forest transect at Bosque Protector San Lorenzo, Panama

Table D.2 ...... 245 Environmental and isotopic data from vertical forest transect at the Smithsonian Ecological Research Center, Edgewater, MD

Table D.3 ...... 247 Biomass and annual litter flux data from Pasoh Forest Preserve Malaysia and carbon isotope data from San Lorenzo used in Monte Carlo bulk leaf isotope resampling model

Table D.4 ...... 248 Biomass and annual litter flux data from the Smithsonian Ecological Research Center, Edgewater, MD used in Monte Carlo resampling model

Table D.5 ...... 249 Carbon isotope measurements for fossil leaves from the Guaduas Formation

Table D.6 ...... 251 Carbon isotope measurements for fossil leaves from the Cerrejón Formation, Site 315

Table D.7 ...... 253 Carbon isotope measurements for fossil leaves from the Cerrejón Formation, Site 318

Table D.8 ...... 255 Alkane carbon isotope measurements for fossil leaves from the Guaduas Formation

Table D.9 ...... 256 Alkane carbon isotope measurements for fossil leaves from the Cerrejón Formation, Site 315

Table D.10 ...... 257 Alkane carbon isotope measurements for fossil leaves from the Cerrejón Formation, Site 318

xiv Table D.11 ...... 258 Alkane carbon isotope measurements for sediments and rock samples from the Guaduas and Cerrejón Formations

Table D.12 ...... 259 Fossils with no quantifiable alkane content

Table D.13 ...... 260 Alkane abundance measurements for fossil leaves from the Guaduas Formation

Table D.14 ...... 261 Alkane abundance measurements for fossil leaves from the Cerrejón Formation, Site 315

Table D.15 ...... 262 Alkane abundance measurements for fossil leaves from the Cerrejón Formation, Site 318

Table D.16 ...... 263 Alkane abundance measurements for sediments and rock samples from the Guaduas and Cerrejón Formations

Table D.17 ...... 264 Lipid biosynthetic fractionation values calculated from bulk fossil and alkane carbon isotope data leaves from Cerrejón and Guaduas Formations

Table D.18 ...... 265 Data and results of mixing model for predicting the relative alkane 13 abundance and δ Clipid expression for alkanes in bulk rock at Cerrejón Formation fossil assemblages

Table D.19 ...... 269 Alkane carbon isotope measurements for leaves from the San Lorenzo forest

Table D.20 ...... 271 Alkane abundance measurements for leaves from the San Lorenzo forest

Table D.21 ...... 273 Lipid biosynthetic fractionation values calculated from leaf and alkane carbon isotope data leaves for San Lorenzo forest leaves

xv Table D.22 ...... 275 Literature values for alkane abundance and chain-length distribution. Data arranged by broad taxonomic groups

Table D.23 ...... 283 13 Total organic carbon and bulk δ CTOC values for samples taken from Mar-2x core. Samples span time periods before, during and after the PETM interval

Table D.24 ...... 285 13 Compound specific δ Clipid values for samples alkanes extracted from Mar-2x core. Samples span time periods before, during and after the PETM interval

Table D.25 ...... 287 Alkane abundance and chain-length distribution for samples taken from Mar-2x core. Samples span time periods before, during and after the PETM interval

xvi Foremost I would like to thank Kate Freeman for taking in a refugee graduate student and providing such valuable guidance and mentorship. I would also like to thank Scott Wing for helping me make my way to Penn State and for having the courage to take a cold call from an unknown graduate student and lend such helpful advice. It has been a pleasure working with both of you! I would also like to thank my committee and collaborators for their commitment to my work. I would like to especially acknowledge someone most of you don’t know, my organic chemistry professor, Jamey Anderson, from Santa Monica

College. He was the first person to offer me a research opportunity and talked me out of nursing school and into chemistry and graduate school. I’m undoubtedly financially much poorer for having taken his advice, but intellectually much richer. Thank you, all of you!

This work would not have been possible without the assistance of the many great scientists at the Smithsonian Tropical Research Center. I would especially like to thank

Fabiany Herrera and Oris Rodriguez for their help during my sampling trips to Panama and Carlos Jaramillo for hosting me and helping me frame this work. I have had great lab mates over the years as well and would especially like to thank Aaron Diefendorf and

Kevin Mueller for the companionship and collaboration we have enjoyed – I hope we can do it again some time! I have also had wonderful, supportive friends – Leah Schneider,

Anna Henderson, and Soumaya Belmecheri - that share my interests (both scientific and not) who have helped me maintain composure over the years and I’d like to thank them.

Most of all I’d like to thank my family from the bottom of my heart for being helpful and trying hard to understand my “academic masochism”. Thanks so much to my parents,

xvii Bob and Betty, for getting me a library card early in life and always encouraging my curiosity. Thanks to my brother Iain for being a great pal, working visits around my silly schedule and recommending great lab work music/noise. To my partner of 19 years, Eric

Church, you are such a hero to me and have been such a great encouragement. Thank you for taking midnight phone calls and never letting me quit graduate school - even though I threatened to at least once a week. I can’t wait to live with you again! I hope that if I ever need ransom money I can still count on all you guys. Much love to you all.

xviii

Recordar el futuro. Imaginar el pasado.

Remember the Future. Imagine the Past.

The future has no more powerful anchor than the past because the past is the only certifiable future we have: the past is the only proof we now have the future did, in effect, once exist.

Carlos Fuentes, The New Salmagundi Reader, 1996

xix Chapter 1. Introduction

1.1. Introduction

The dense, closed-canopy forests of the moist tropics are among the most diverse terrestrial ecosystems. Alpha diversity (number of species in an ecosystem or “species richness”) in closed-canopy tropical forests can reach up to 300 tree species per hectare

(Phillips et al., 1994; Turner, 2001; Wright, 2002). Species co-exist in a wide spectrum of niches. These niches can be defined horizontally, across the landscape, by topographic and edaphic (soil-related) variations (Connell & Grubb, 1996; Clark et al., 1998). More importantly, niches are defined vertically by the canopy structure of the forest. The wide variety of tree heights combined with extreme gradients in light and humidity from canopy top to forest floor provide a diversity of microhabitats and light environments that allow plants to specialize into different strata of the forest as they mature (Augspurger,

1984; Kohyama, 1993, Myneni et al., 2007).

The geologic history of forest structure and canopy closure is not well understood. Fossil assemblages seldom reflect tree spacing, size, or branching patterns (Tiffney, 1984; Davis et al., 2005). Adaptations specific to closed-canopy forests – such as large, fleshy fruits or branchless boles – rarely preserve (Bruun & Ten Brink, 2008). The best preserved and most abundant plant fossils are leaves and leaf fragments, however leaf morphology alone does not indicate canopy conditions. This study focuses on leaf biochemical signatures that have the potential to record light, humidity, and isotopic gradients,

1 preserved in geologic archives, and ultimately serve as proxies of canopy density. Study results have been applied to analysis of Neotropical leaf fossils. This introductory chapter explores the previous research that informs this topic and the specific goals of our work.

1.2. Climatic feedback in closed-canopy forests

Closed-canopy forests affect regional and global climate. Evapotranspiration in tropical rainforests lowers albedo and increases surface moisture availability, which in turn regulates global temperature, hydrologic cycling, and atmospheric circulation (Bastable et al., 1993; Betts, 1999; Boyce et al., 2010). Closed-canopy tropical forests thrive in year-round warmth (> 20° C) and high precipitation (> 2000 mm/year) regimes (Turner,

2001). Tropical rainforests are only ~12% of the Earth's surface but account for 45% of carbon in terrestrial biomass (IPCC, 2000; Grace et al., 2001; Malhi et al., 2002). This makes the origin and evolution of closed-canopy forests a topic of interest to both climatologists and paleo-ecologists alike.

1.3. Reconstruction of ancient canopied forests from fossils

The rise of closed-canopy forests is often linked with the widespread distribution of angiosperms in the Paleocene, but the location and age of the first forest canopies is disputed (Germeraad et al., 1968; Lappartient, 1981; Upchurch & Wolfe, 1987; Spicer et al., 1993; Morley, 2000; Johnson & Ellis, 2002; Burnham & Johnson, 2004; Davis et al.,

2 2005; Jaramillo et al., 2010). Reconstructions of canopy in the geologic past typically rely on physiognomic features of fossil leaves as an indicator of climate or light environment, or on identification of taxa associated with extant closed-canopy forests

(Johnson & Ellis, 2002; Davis et al., 2005; Wing et al., 2009). Positive identification of stand architecture from fossil leaves is difficult since morphological differences can be very subtle and require exceptional preservation (Burnham et al., 1992; Ellis et al., 1999).

Further, fossil assemblages include extinct taxa that are not necessarily analogous to modern forests and lack extant taxa that we have used to shape our interpretation of floristic components.

1.4. The origin and evolution of the closed-canopy forest in the Neotropics

Neotropical forests of Central and South America have developed different family-level dominance than those of Africa and Asia since the separation of Gondwana. The

Cretaceous plant record of the Neotropics is best known from pollen assemblages, most of which are dominated by gymnosperms. Angiosperm pollen is first reported from the

Late Aptian but cannot be allied to modern taxa (Herngreen & Duenas Jimenez, 1990).

Albian pollen deposits can be allied with the Arecaceae in many cases but is dominated by dicotyledenous angiosperm pollen (Burnham & Johnson, 2004). Although shade tolerance is not unique to the angiosperms, these taxa represent 80-90% of species in modern, closed-canopy tropical forests, which suggest the Late Cretaceous increase in angiosperms could be associated with greater canopy closure (Burnham & Johnson,

2004; Valladares & Niinemets, 2008; Jaramillo et al., 2010).

3

Paleoflora from the Paleocene-age Cerrejón Formation in Colombia contain all the floristic elements associated with closed-canopy forests, and reconstructed climatic parameters suggest a tropical rainforest biome (Wing et al., 2004; Herrera, 2008a;

Herrera, 2008b; Wing et al., 2009). The diversity of families in the Cerrejón fossil assemblages is very low, though, when compared with modern closed-canopy environments (Wing et al., 2009). This lack of diversity could be interpreted as a slow recovery from the end-Cretaceous extinction or could reflect the bias in litter from closed-canopy forests towards leaves from the upper canopy (Ferguson, 1985; la Parra et al., 2007). Litter from closed-canopy forests contains leaves mostly from species that dominate the canopy and does not reflect the full diversity of the forest. With a clearer biogeochemical indication of canopy closure it would be easier to understand low diversity in paleofloral assemblages from closed-canopy forests. In addition to informing interpretations of fossil leaf assemblages, a biomarker approach could also open up geologic records that lack well-preserved fossils for interpretation of the canopy structure.

1.5. Biochemical features of canopy closure: The canopy effect

One of the most defining features of the closed-canopy forest is a strong vertical light gradient. As little as 1% of available photosynthetic photon flux density reaches plants on the forest floor (Chazdon & Fetcher, 1984; Niinemets, 2007; Santiago & Wright, 2007).

This extreme gradient in light leads to a greater than 100% increase in photosynthetic rate

4 from forest floor to canopy top (Meir et al., 2002). Along this photosynthetic gradient, plants express strong physiological trends in anatomical and chemical leaf traits (Medina

& Minchin, 1980; Francy et al., 1985; Leaney et al., 1985; Farquhar et al., 1989;

Niinemets et al., 1999; Beerling & Royer, 2002; Meir et al., 2002; Poorter et al., 2006;

Boyce, 2008). Leaves in the understory can have lifespans ten times longer than canopy leaves (Reich et al., 1991).

13 13 The C-abundance in bulk leaf tissue (expressed as δ Cleaf – see section 1.11 for more details on isotopic notation) declines downward in the canopy. This “canopy effect” is a long-recognized indicator of canopy conditions (Vogel, 1978). Three primary mechanisms contribute to the canopy effect. First, due to high respiration by soil biota

13 and poor atmospheric mixing, the CO2 in an understory is often C depleted; this isotopic depletion can be recorded in leaf tissue during photosynthesis (Medina &

Minchin, 1980; Da Silveira et al., 1989; Brooks et al., 1997). Second, understory humidity is higher, permitting stomata to remain open without loss of leaf-water, reducing the difference between leaf interior and atmospheric CO2 concentrations and

13 resulting in a fuller expression of enzymatic CO2 discrimination (Δleaf; Farquhar et al.,

1989; Madhavan et al., 1991; also, see section 1.11 for details on calculation). Third, elevated carbon assimilation rates in the canopy top lead to greater draw-down of internal

13 CO2 and much less C discrimination when compared to leaves lower in the canopy

(Francey et al., 1985; Ehleringer et al., 1986; Zimmerman & Ehleringer, 1990; Lockheart et al., 1997; Niinemets et al., 1999).

5 13 δ Cleaf values indicative of canopy closure are potentially preserved in fossil leaf material, although this application has not been explored. Rather, measurements of

13 δ Cleaf from fossil angiosperm leaves have been used to make inferences about carbon cycle perturbations, molecular taphonomy, and chemotaxonomy and assume that leaf fossils represent the entire forests and are not influenced by ecosystem structure (Otto et al., 1994; Huang et al., 1995; Gupta et al., 2007; Zanetti et al., 2007; Xiao et al., 2013).

These published studies indicate biomarker compounds are more likely to be found in fossils rather than sediments, how they differ from modern relatives, and how stomata anatomy and carbon metabolism have responded to changes in atmospheric CO2.

1.6. Isotopic reconstruction of ancient ecosystems

13 The δ CTOC values of organic matter found in soils and sediments has been used to infer a variety of climate conditions, as well as ecological features such as the phylogenetic dominance of plants on the landscape (Bowen et al., 2004; Magioncalda et al., 2004;

Wing et al., 2005; Schouten et al., 2007; Secord et al., 2008; Magill et al., 2013). The use of terrestrial organic carbon in geologic archives for the purpose of plant community reconstruction is problematic since these substrates integrate all preserved carbon– not just plant biomass. Further, reconstructions of plant communities are generally based on comparisons of just one modern plant organ (the leaf) and rarely consider the contribution of other plant organs (roots, bark, stems, etc) as well as associated epiphytic and mycorrhizal fungal communities into the soil organic carbon pool (Grocke, 2002;

Bowling et al., 2008; Mueller et al., 2012). For this reason many workers now advocate

6 plant-specific biomarkers as a better proxy of carbon from ancient plants (Bird et al.,

1995; Briggs et al., 2000; Lockheart et al., 2000; van Aarssen et al., 2000; Pancost &

Boot, 2004; Smith & Freeman, 2006; Smith et al., 2007; Polissar et al., 2009; Jaramillo et al., 2010).

1.7. The advantages of plant-specific biomarkers in ecosystem reconstruction

High-molecular weight n-alkanes with odd numbers of carbon atoms (C25 – C35), odd- originate in leaf cuticle and are common in the rock record, often independent of plant macrofossil remains (Eglinton & Hamilton, 1963; Kolattukudy, 1976; Cranwell, 1981).

Relative to biomass, n-alkanes are exceptionally resistant to chemical change and microbial alteration (Eglinton & Logan, 1991; van Bergen et al., 1995).

Long-chain n-alkanes extracted from leaves are more depleted in 13C than bulk leaf material (Collister et al., 1994); this isotopic depletion can range widely (4-12‰)

(Chikaraishi et al., 2004; Diefendorf et al., 2011; Diefendorf et al.,2012). Extracted cuticle lipids have long been used as paleobiological and paleo-environmental isotopic proxies (Hayes et al., 1989; Rieley et al., 1991; Freeman & Colarusso, 2001; Pagani et al., 2006; Smith et al., 2007; Pedentchouk et al., 2008; Tipple & Pagani, 2010). The extent to which this fractionation is influenced by environmental gradients or the isotopic canopy effect has not yet been explored.

7 1.8. Ecophysiology of leaf alkane production

n-Alkanes are a major wax component of plant cuticle along with alcohols, ketones, and fatty acids. Broadly, cuticle refers to the coating found on aerial plant organs. From a biomass flux standpoint leaves are the major source of alkanes in sediments. The cuticular membrane can range from just a few nanometers thick to tens of micrometers and varies widely between and within taxonomic groups (de Vries, 1968; Sargent, 1976;

Heide-Jorgensen, 1978; Heide-Jorgensen, 1980; Jeffree, 2006). The concentration of alkanes in cuticle is lower in plants subjected to drought conditions (Baker, 1974). Also, experiments have found that increases in temperature can cause an increase in the wax amount but a decrease in the relative concentration of alkanes in those waxes (Jenks et al., 2001).

Long-lived leaves in the deeply shaded understory contain far more defense compounds against herbivory than their high-light counterparts (Coley, 1988). Recent work indicates that leaf life span is associated with overall higher abundances of both foliar n-alkanes and terpenoid compounds in evergreen specimens (Diefendorf et al., 2011; Diefendorf et al,. 2012). The effects of light regime on both n-alkane production and chain-length distribution are largely unexplored. Greater allocation of carbon to defense compounds in longer-lived leaves by understory plants could possibly influence the observed isotopic fractionation (εlipid, see section 1.11 for details on this calculation) between whole leaf tissue and individual lipids such as n-alkanes.

8 1.9. Factors affecting preservation of biomarkers

Reconstruction of a paleoenvironment from fossil plant material requires not only an understanding of the controls on production and distribution of plant biomass and lipid biomarkers within ecosystems, but also an understanding of the transport, preservation, and depositional environments of fossil and lipid substrates. Alkanes and plant macrofossils share the same preferred preservation environment. In particular, compression leaf fossils that retain organic matter are preserved almost exclusively in aqueous anoxic settings characterized by fine-grained sediments and humic acids

(Stewart & Rothwell, 1993). Similarly, alkanes also preserve better under anoxic conditions and are found in conjunction with mineral surfaces or sorbed on to mineral, hence clays and other fine-grained sediments are an excellent source of abundant alkanes

(Hedges & Keil, 1995).

n-Alkanes enter sedimentary systems by a variety of transport mechanisms. Alkanes can be deposited by wind either as aerosols or as adsorbed components in dust (Simoneit et al., 1977; Gagosian et al., 1987). Alkanes are also transported into marine depositional environments along deltaic or continental shelves as riverine runoff (Hedges & Keil,

1995; Hedges et al., 1997; Schlunz & Schneider, 2000). Plant-derived n-alkanes can enter waterways either from degraded vegetation in the water column or from soils that erode into the water (Meyers & Ishiwateri; 1993; Goni et al., 1997; Seki et al., 2010). Leaf wax n-alkanes can enter soils either by wind, sand, or rainfall ablation (Colina-Tejada et al.,

1997). Most alkanes in soils are released though degradation of leaf litter.

9

1.10. Factors affecting organic matter preservation in leaf fossils

Biochemical analysis of plant fossils should consider the nature of fossilization and the biases any modifications or decomposition that may be inherent in the process. The path from the plant to the rock record can be summarized as (1) abscission and accumulation,

(2) transport and mixing, and (3) burial (Rex & Chaloner, 1983; Burnham et al., 1992;

Stewart & Rothwell, 1993). Taphonomic forces can affect the quantity and condition of leaves preserved in the fossil record. Plant fossil assemblages are not just a summary of the living flora local to the accumulation point or a simple integration of synchronously senesced leaves.

Leaf assemblages are dominated by the abundant species on a landscape and do not necessarily record total species richness or the distribution of taxa (Burnham et al., 1992).

Unequal decay processes in this litterfall will also alter the expressed distribution of leaves in paleofloral assemblages. Decay rates can be affected by soil alkalinity, moisture, and mineral composition but most importantly by the chemical characteristics of the leaf itself, which are dictated by taxonomy and the growth conditions of an individual leaf (Swift et al., 1979; Hill & Gibson, 1986; Ferguson, 1985; Spicer, 1989; see Mueller et al., 2013). Upon abscission, leaf traits such as weight, size, shape, and petiole length can all affect leaf-fall and the distance a leaf is transported (Ferguson,

1985). These factors can bias the fossil record towards certain leaf forms. Transport

10 processes can or may result in a mixing of leaves from a wider area than represented by autochthonous deposits (Spicer, 1981; Spicer & Wolfe, 1989).

Alkanes are best preserved in the same rapidly formed, anoxic sedimentary deposits as fossil leaves (Briggs et al., 2000). Biomarker preservation in fossil leaf cuticle has only been explored in one locality (Huang et al., 1995). More work is necessary in order to understand the conditions where biochemical analyses of fossils are best suited

1.11. Isotopic notation and calculation of isotope fractionation

All carbon isotope values are reported in delta (δ) notation relative to the Vienna Pee Dee

Belemnite (VPDB) standard such that

13 13 13 13 13 12 δ C = [( Rsample/ RVPDB) – 1] and R = C/ C

δ values reflect the deviation of a sample from a standard (in this case the Vienna Pee

Dee Belemnite) in parts per thousand (‰ or “permil”). These values can describe very small variations in the relative abundances of rare isotopes in organic matter.

Apparent isotope fractionation is an estimate measurable mass difference between a reactant and a product due to a specific reaction. This metric is generally defined as

13 13 αa-b = Ra/ Rb

11 but values from this calculation are difficult to use in practice since the variation is expressed in the 4th or 5th decimal place. In this work, we use specific notation to refer to the isotopic fractionation that occurs during photosynthesis (Δleaf), when atmospheric CO2 is fixed into leaf tissue (Farquhar et al., 1989)

13 13 13 ∆leaf = (δ Catm - δ Cleaf)/(1 + δ Cleaf / 1000)

We use a different notation to refer to lipid biosynthetic fractionation (εlipid), the difference between leaf tissue and the lipids that arise within tissues (Hayes et al., 2001).

13 13 εlipid = [(δ Clipid + 1000) / (δ Cleaf + 1000) - 1] x 1000

Both equations are algebraically equivalent, though reactants and products are reversed.

The different fractionation notations reflect differences in use by different disciplines:

Δleaf is a common unit in ecological literature while εlipid is more common in the geology literature.

12 1.12. Study objectives and chapter outlines

13 This dissertation explores the vertical variations in δ C values for leaves and leaf lipids in closed-canopy forests. We concentrate on the areally constrained, intra-organismal

13 δ C variations that arise in response to canopy closure. We then apply our findings to fossil archives in order to understand the history of canopy closure in the Neotropics

13 Chapter 2: Closed-canopy forests exhibit a wide range of δ Cleaf values, whereas more

13 open ecosystems express a narrow range of δ Cleaf values. How many fossil

leaves would need to be analyzed from a fossil assemblage in order to capture

13 a diagnostic range of δ Cleaf values that would indicate if the canopy was

open or closed?

13 Chapter 3: Organic carbon is preserved in fossil leaves. The range of δ Cleaf values in a

collection of leaves should reflect the extent of canopy coverage. Will the

13 range of δ Cleaf values from leaves in a paleoflora confirm the floristic

13 evidence of canopy closure? Will the range of δ Cleaf values differ among

fossil leaf assemblages enough to identify open or forest edge assemblages?

Chapter 4: The isotopic composition of leaf-derived sedimentary carbon and preserved

alkanes is often used to infer ecologic or environmental conditions. Generally,

the extent of isotope fractionation between carbon pools is assumed to be the

13 13 13 same in the past as it is in extant organisms. We compare δ Cleaf and δ Clipid

values from fossil and modern leaves lto explore fractionation factors for

important plant families of the Neotropics. How similar are the ∆leaf and εlipid

values expressed in leaves from fossil and extant trees? Has carbon

metabolism changed appreciably in similar plant families since the Eocene?

Chapter 5: Fossils preserve leaf carbon and alkanes but it is not known if distinctive

patterns of alkane abundance or isotopic composition or if sediments preserve

lipid characteristics of the forest stand. Are patterns of isotopic expression

preserved in alkanes from fossil leaves when compared to modern relatives?

What sedimentary and depositional conditions promote biomarker

preservation in paleoflora? Can biomarkers and their isotopes be used to

identify biochemical features of forest structure in a paleoflora? How well

does the alkane complement of a sediment represent the assemblage of alkane

source leaves?

Chapter 6: The isotopic composition and chain-length distribution of alkanes is often

used to infer environmental conditions or phylogentic composition for plant

communities in the geologic past but the intra-organismal variations in alkane

expression within a canopy on small spatial scales have not been explored.

What are the vertical patterns of alkane production in closed-canopy

ecosystems? Can these variations in alkane production be defined

14 environmentally or taxonomically? Is there an evolutionary influence on

alkane production?

1.13. Publications arising from this dissertation

Each chapter constitutes an individual publishable work. Chapters provide more detail than anticipated publication and are meant to provide a cohesive theme to the dissertation. Current and anticipated publications arising from this work include:

Chapter 2: “Isotopic characteristics of canopies in simulated leaf assemblages” Accepted

with revisions in 2013 Geochimica et Cosmochimica Acta Heather V. Graham

with co-authors Mark Patzkowsky, Scott L. Wing, Geoffrey G. Parker,

Marilyn Fogel, and Katherine H. Freeman.

Chapter 3: “Biogeochemical confirmation of the first closed-canopy Neotropical forest in

the fossil record: A novel application of isotopic methods”, will be submitted

to Geology by Heather V. Graham with co-authors Scott L. Wing, Fabiany

Herrera, and Katherine H. Freeman.

Chapter 4: “Plant waxes and fossil leaves from Cretaceous and Paleocene Neotropics: a

Comparison of Modern and Ancient Fractionation Factors”, will be submitted

to Organic Geochemistry by Heather V. Graham with co-author Katherine H.

Freeman.

15 Chapter 5: “Simulated and observed alkane distribution and isotope composition in

sediments using biogeochemical analysis of fossil leaves”, will be submitted

to Organic Geochemistry by Heather V. Graham with co-author Katherine H.

Freeman.

Chapter 6: “Alkane production in angiosperms: Understanding the influences of

phylogeny and ecophysiology”, will be submitted to New Phytologist by

Heather V. Graham with with co-authors Scott L. Wing, Anna Henderson and

Katherine H. Freeman.

16 1.14. References

Augspurger C.K. (1984) Light requirements of neotropical tree seedlings: a comparative study of growth an survival. J. Ecol. 72, 777-795.

Baker E.A. (1974) The influence of environment on leaf wax development in Brassica oleraceae var. gemmifera. New Phytologist 73, 955-966.

Bastable H.G., Shuttleworth J., Dallarosa R.L.G., Fisch G. and Nobre C.A. (1993) Observations of climate, albedo, and surface radiation over cleared and undisturbed Amazonian forest. Int. J. of Climatol. 13, 783-796.

Beerling D.J. and Royer D.L. (2002) Fossil plants as indicators of the Phanerozoic global carbon cycle. Ann. Rev. Earth Planet. Sci. 30, 527-556.

Betts R.A. (1999) Self-beneficial effects of vegetation on climate in an Ocean- Atmosphere General Circulation model. Geophys. Res. Lett. 26, 1457-1460.

Bird M.I., Summons R.E., Gagan M.K., Roksandic Z., Dowling L., Head J., Keith Fifield L., Cresswell R.G. and Johnson D.P. (1995) Terrestrial vegetation change inferred from n-alkane δ13C analysis in the marine environment. Geochim Cosmochim Acta 59, 2853-2857.

Bowen G.J., Beerling D.J., Koch P.L., Zachos J.C. and Quattlebaum T. (2004) A humid climate state during the Palaeocene/Eocene thermal maximum. Nature 432, 495- 499.

Bowling D.R., Patakie D.E. and Ehleringer J.T. (2008) Carbon isotopes in terrestrial ecosystem pools and CO2 fluxes. New Phytologist 178, 24-40.

Boyce C.K. (2008) The fossil record of plant physiology and development - what leaves can tell us. Paleontolog. Soc. Papers 14, 133-146.

Boyce C.K., Lee J-E., Field T.S., Brodribb, T.J. and Zwieniecki M.A. (2010) Angiosperms helped put the rain in the rainforests: the impact of plant physiological evolution on tropical biodiversity. Ann. MO Bot. Garden 97, 527-5 40.

Briggs D. E. G., Evershed R. P. and Lockheart M. J. (2000) The biomolecular paleontology of continental fossils. Paleobiol. 26, 169–193.

Brooks J.R., Flanagan L.B., Varney G.T. and Ehleringer J.R. (1997) Vertical gradients in photosynthetic gas exchange characteristics and refixation of respired CO2 within boreal forest canopies. Tree Physiol. 17, 1-12.

17 Bruun H.H. and Ten Brink D.J. 2008. Recruitment advantage of large seeds is greater in shaded habitats. Ecoscience 15 498-507.

Burnham R.J., Wing S.L. and Parker G.G. (1992) The reflection of deciduous forest communities in leaf litter: Implications for autochthonous litter assemblages from the fossil record. Paleobiol. 18, 30-49.

Burnham R.J. and Johnson K.R. (2004) South American palaeobotany and the origins of neotropical rainforests. Proc. R. Soc. London, Ser. B 359, 1595-1610.

Chazdon, R.L. and Fetcher N. (1984) Photosynthetic light environments in a lowland tropical rainforest in Costa Rica. J. Ecol. 72, 553-564.

Chikaraishi Y. and Naraoka H. (2003) Compound-specific δD-δ13C analyses of n-alkanes extracted from terrestrial and aquatic plants. Phytochem. 63, 361-371.

Clark D.B., Clark D.A. and Read J.M. (1998) Edaphic variation and the mesoscale distribution of tree species in a neotropical rainforest. J. Ecol. 86, 101-112.

Colina-Tejada A. Amblès A. and Jambu P. (1996) Nature and origin of soluble lipids shed into the soil by rainwater leavhing a forest cover of Pinus maritime sp. Eur. J. Soil Sci. 47, 637-643.

Collister J.W., Rieley G., Stern B., Eglinton G. and Fry B. (1994) Compound-specific δ13C analyses of leaf lipids from plants with differing carbon dioxide metabolisms. Org. Geochem. 3, 619-627.

Coomes D.A. Grubb P.J. (1996) Amazonian caatinga and related communities at La Esmeralda, Venezuela: forest structure, physiognomy and floristics, and control by soil factors. Vegetatio 122, 167-191.

Cranwell P.A. (1981) Diagenesis of free and bound lipids in terrestrial detritus deposited in a lacustrine sediment. Org. Geochem. 3(3), 79-89.

Davis C.C., Webb C.O., Wurdack K.J., Jaramillo C.A. and Donoghue M.J. (2005) Explosive radiation of Malpighiales supports a mid-Cretaceous origin of modern tropical rainforests. American Naturalist 165, E36-E65.

De la Parra F., Jaramillo C.A., and Dilcher D.L. (2007) Paleoecological changes of spore producing plants through the Cretaceous-Paleocene boundary in Colombia. Palynol. 32, 258-259.

De Vries H.A.M. (1968) Development of the structure of the normal, smooth cuticle of the apple ‘Golden delicious’. Acta Bot. Neerlandica 17, 229-241.

18 Diefendorf A.F., Mueller K.E., Wing S.L., Koch P.L. and Freeman K.H. (2010) Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proc. Natl. Acad. Sci. 107, 5738-5743.

Diefendorf A.F., Freeman K.H., Wing S.L. and Graham H.V. (2011) Production of n- alkyl lipids in living plants and implications for the geologic past. Geochim. Cosmochim. Acta 75, 7472-7485.

Eglinton G. and Hamilton R.J. (1967) Leaf epicuticular waxes. Science 156, 1322-1335.

Eglinton G. and Logan G.A. (1991) Molecular preservation. Phil. Trans. Roy. Soc. London B. 333, 315-328.

Ehleringer J.R., Field C.B., Lin Z. and Kuo C. (1986) Leaf carbon isotope and mineral composition in subtropical plants along irradiance cline. Oecologia 70, 520-526.

Ellis B., Daly D.C., Hickey L.J., Johnson K.R., Mitchell J.D., Wilf, P. and Wing S.L. (2009) Manual of Leaf Architecture. Cornell University Press, Ithaca, USA.

Farquhar G.D., Ehleringer J.R. and Hubick K.T. (1989) Carbon isotope discrimination and photosynthesis. Ann. Rev. Plant Physiol Plant Mol. Biol. 40, 503-537.

Ferguson D.K (1985) The origin of leaf-assemblages – New light on an old problem. Rev. Palaeobot. Palynol. 46, 117-188.

Field C.B., Behrenfeld M.J., Randerson J.T. and Falkowski P. (1998) Primary production of the biosphere: integrating terrestrial and oceanic components. Science 281, 237-240.

Francey R.J., Gifford R.M., Sharkey T.D. and Weir B. (1985) Physiological influences on carbon isotope discrimination in huon pine (Lagarostrobos franklinii). Oecologia 66, 211-218.

Freeman K.H. and Colarusso L.A. (2001) Molecular and isotopic records of C4 grassland expansion in the late Miocene. Geochim. Cosmochim. Acta 65, 1439-1454.

Gagosian R.B. and Peltzer E.T. (1986) The importance of atmospheric input of terrestrial organic material to deep-sea sediments. Org. Geochem. 10, 661-669.

Germeraad J.H., Hopping C.A. and Muller J. (1968) Palynology of Tertiary sediments of tropical areas. Rev. Palaeobot. Palynol. 6, 189-348.

Grace J., Malhi Y., Higuchi N. and Meir P. (2001) Productivity of tropical rainforests. In Terrestrial global productivity: past, present and future (eds. H. Mooney, J. Roy and B. Saugier). Academic Press, San Diego, USA. pp. 401-426.

19 Goñi M.A., Ruttenberg K.C. and Eglinton T.I. (1997) Source and contribution of terrigenous organic carbon to surface sediments in the Gulf of Mexico. Nature 389, 275-278.

Grocke D.R. (2002) The carbon isotope composition of ancient CO2 based on higher- plant organic matter. Philos. Trans. R. Soc. London, Ser. A 360, 633-658.

Gupta N. S., Briggs D. E. G., Collinson M. E., Evershed R. P., Michels R. and Pancost R.D. (2007) Molecular preservation of plant and insect cuticles from the Oligocene Enspel Formation, Germany: Evidence against derivation of aliphatic polymer from sediment. Org. Geochem. 38, 404–418.

Hedges J.I., Keil R.G. and Benner R. (1997) What happens to terrestrial organic matter in the ocean? Org Geochem. 27, 195-212.

Hedges J.I. and Keil R.G. (1995) Sedimentary organic matter preservation: an assessment and speculative synthesis. Marine Chem. 49, 81-115.

Hayes, J.M. (2001) Fractionation of carbon and hydrogen isotopes in biosynthetic processes. Rev. Mineral. Geochem. 43, 225-277.

Heide-Jørgensen H.S. (1980) Cuticle development and ultrastructure – evidence for a procuticle of high osmium affinity. Planta. 183, 511-519.

Herngreen G.F.W. and Duenas Jimenez H. (1990) Dating of the Cretaceous Une formation Colombia and the relationship with the Albian-Cenomanian African- South American microfloral province. Rev. Palaeobog. Palynol. 66, 345-359.

Herngreen G.F.W., Kedves M., Rovina L.V. and Smirnova S.M. (1996) Cretaceous palynofloral provinces: a review. In Palynology: Principles and applications (eds. J. Jansonius and D.C. McGregor). Association of Stratigraphic Palynologists Foundation, Dallas, USA. pp. 1157-1188.

Herrera F., Jaramillo C., Dilcher D., Wing S.L. and Gomez C. (2008) Fossil Araceae from a Paleocene Neotropical rainforest in Colombia. Am. J. Bot. 95, 1659-1683.

Herrera F., Manchester S.R., Hoot S.B., Wefferling K., Carvalho M. and Jaramillo C. (2011) Phytogeographic implications of fossil endocarps of Menispermaceae from the Paleocene of Colombia. Am. J. Bot. 98, 1-14.

Hill R.S. and Gibson N. (1986) Distribution of potential macrofossils in Lake Dobson, Tasmania. J. Ecol. 74, 373-384.

Huang Y., Lockheart M. J., Collister J. W. and Eglinton G. (1995) Molecular and isotopic biogeochemistry of the Miocene Clarkia Formation: hydrocarbons and alcohols. Org. Geochem. 23, 785–801.

20

IPCC (2000) Land use, land use change and forestry. A special report of the IPCC. Cambridge University Press, Cambridge UK.

Jaramillo C.A., Ochoa D., Contreras L., Pagani M., Carvajal-Ortiz H., Pratt L.M., Krishnan S., Cardona, Romero M., Quiroz L., Rodriguez G., Rueda M.J., de la Parra F., Morón S., Green W., Bayona G., Montes C., Quintero O., Ramirez F., Mora G., Schouten S., Bermudez H., Navarrete R., Parra F., Alvarán M., Osorno J., Crowley J.L., Valencia V. and Vervoort J. (2010) Effects of rapid global warming at the Paleocene-Eocene boundary on neotropical vegetation. Science 330, 957-961.

Jeffree C.E. (2006) The fine structure of the plant cuticle . In: Biology of the Plant Cuticle (eds. M. Riederer and C. Müller). Blackwell Publishing, Oxford, UK. pp. 11-125.

Jenks M.A., Andersen L., Teusink R.S. and Williams M.H. (2001) Leaf cuticular waxes of potted rose cultivars as affected by plant development, drought and paclobutrazol treatments. Physiolog. Plantarum 112, 62-70.

Johnson K.R. and Ellis B. (2002) A tropical rainforest in Colorado 1.4 million years after the Cretaceous-Tertiary boundary. Science 296, 2379-2383.

Kohyama T. (1993) Size-structured tree populations in gap-dynamic forest; the forest architecture hypothesis for the stable coexistence of species. J.Ecol. 81, 131-143.

Kolattukudy P.E. (1976) The Chemistry and Biochemistry of Natural Waxes. Elsevier, Amsterdam.

Leaney F.W., Osmond C.B., Allison G.B., and Zeigler H. (1985) Hydrogen-isotope composition of leaf water in C3 and C4 plants: its relationship to the hydrogen- isotope composition of dry matter. Planta 164, 215-220.

Lockheart M.J., Van Bergen P.F. and Evershed R.P. (1997) Variations in the stable carbon isotope compositions of individual lipids from the leaves of modern angiosperms: implications for the study of higher land plant-derived sedimentary organic matter. Org.Geochem. 26, 137-153.

Madhavan S., Treichel I. and O’Leary M.H. (1991) Effect of relative humidity on carbon isotope fractionation in plants. Botany Acta 104, 292-294.

Magill C.R., Ashley G.M. and Freeman K.H. (2013). Ecosystem variability and early human habitats in eastern Africa. Proc. Nat. Acad. Sci. 4, 1167-1174.

21 Magioncalda R., Dupuis C., Smith T., Sterubaut E. and Gingerich P.D. (2004) Paleocene- Eocene carbon isotope excursion in organic carbon and pedogenic carbonate: direct comparison in a continental stratigraphic section Geol. 32, 553-556.

Malhi Y., Meir P. and Brown S. (2002) Forests, carbon and global climate. Philos. Trans. R. Soc. London, Ser. B 354, 1739-1748.

Medina E. and Minchin P. (1980) Stratification of δ13C values of leaves in Amazonian rainforests. Oecologia 45, 377-378.

Meir P., Kruijt B., Broadmeadow M., Barbosa E., Kull O., Carswell F., Nobre A. and Jarvis P.G. (2002) Acclimation of photosynthetic capacity to irradiance in tree canopies in relation to leaf nitrogen concentration and leaf mass per unit area. Plant Cell Env. 25, 343-357.

Meyers P.A. and Ishiwatari R. (199e) Lacustrine organic geochemistry – an overview of indicators of organic matter sources and diagenesis in lake sediments. Org Geochem 20, 867-900.

Monteillet J. and Lappartient J. (1981) Fruits et graines du Cretace-Superieur des Carrieres de Paki (Senegal). Rev. Palaeobot. Palynol. 34, 331-344.

Morley R.J. (2000) Origin and Evolution of Tropical Rainforests. John Wiley & Sons, Chichester UK.

Mueller K.E., Polissar P.J., Oleksyn J. and Freeman K.H. (2012) Differentiating temperate tree species and their organs using lipid biomarkers in leaves, roots and soil. Org. Geochem. 52, 130-141.

Myneni R.B., Yang W., Nemani R.R., Huete A.R., Dickinson R.E. and Knyazikhin Y. (2007) Large seasonal swings in leaf area of Amazon rainforests. Proc. Natl. Acad. Sci. 104, 4820-4823.

Niinemets Ü., Kull O. and Tenhunen J.D. (1999) Variability in leaf morphology and chemical composition as a function of canopy light environment in coexisting deciduous trees. Int. J. Plant Sci. 160, 837-848.

Niinemets Ü. (2007) Photosynthesis and resource distribution through plant canopies. Plant Cell Env. 30, 1052-1071

Otto A., Simoneit B. R. T. and Rember W. C. (2005) Conifer and angiosperm biomarkers in clay sediments and fossil plants from the Miocene Clarkia Formation, Idaho, USA. Org. Geochem. 36, 907–922.

Pagani M., Pedentchouk N., Huber M., Sluijs A., Schouten S., Brinkhuis H., Sinninghe Damste J.S., Dickens G.R. and the IODP Expedition 302 Expedition Scientists.

22 (2006) Arctic hydrology during global warming at the Palaeocene-Eocene thermal maximum. Nature 442, 671-675.

Pancost R.D. and Boot C.S. (2004) The palaeoclimatic utility of terrestrial biomarkers in marine sediments. Marine Chem. 92, 239-261.

Pedentchouk N., Sumner W., Tipple B. and Pagani M. (2008) δ13C and δD compositions of n-alkanes from modern angiosperms and conifers: an experimental set up in central Washington State, USA. Org. Geochem. 39, 1066-1071.

Phillips O.L., Hall P., Gentry A.H., Sawyer S.A. and Vasquez R. (1994) Dynamics and species richness of tropical rainforests. Proc. Natl. Acad. Sci. 91, 2805-2809.

Polissar P.J., Freeman K.H., Rowley D., McInerney F.A. and Currie B. (2009) Paleoaltimetry of the Tibetean Plateau from D/H ratios of lipid biomarkers. Earth Planet. Sci. Lett. 287, 64-76.

Poorter H., Pepin S., Rijkers T., de Jong Y., Evans J.R. and Körner C. (2006) Construction costs, chemical composition and payback time of high- and low- irradiance leaves. J. Exp. Bot. 57, 355-371.

Reich P.B., Uhl C., Walters M.B. and Ellsworth D.S. (1991) Leaf lifespan as a determinant of leaf structure and function among 23 Amazonian tree species. Oecologia 86, 16-24.

Rex G.M, and Chaloner W.G. (1983) The experimental formation of plant compression fossils. Paleontol. 26(2), 231-252.

Rieley G., Collier R.J., Jones D.M., and Eglinton G. (1991) The biogeochemistry of Ellesmere Lake, UK: I. Source correlation of leaf wax inputs to the sedimentary lipid record. Org. Geochem. 17, 901-912.

Santiago L.S. and Wright S.J. (2007) Leaf functional traits of tropical forest plants in relation to growth form. Func. Ecol. 21,19-27.

Sargent C. (1976) In situ assembly of cuticular wax. Planta 129, 123-126.

Schlünz B. and Schneider R.R. (2000) Transport of terrestrial organic carbon to the oceans by rivers: re-estimating flux- and burial-rates. Int. J. Earth Sci. 88, 599- 606.

Schouten S., Woltering M., Rijpstra W.I.C., Sluijs A., Brinkhuis H. and Sinninghe Damste J.S. (2007) The Paleocene-Eocene carbon isotope excursion in higher plant organic matter: Differential fractionation of angiosperms and conifers in the Arctic. Earth Planet. Sci. Lett. 258, 581-592.

23 Secord R., Wing S.L. and Chew A. (2008) Stable isotopes in early Eocene mammals as indicators of forest canopy structure and resource partitioning. Paleobiol. 34,282- 300.

Seki O. Nakatsuka T. Shibata H. and Kawamura K. (2010) A compound-specific n- alkane δ13C and δD approach for assessing source and delivery processes of terrestrial organic matter within a forested watershed in northern Japan. Geochim. Cosmochim. Acta. 74, 599-613.

Simeoneit B.R.T. (1997) Compound-specific carbon isotope analyses of individual long- chain alkanes and alkanoic acids in Harmatten aerosols. Atmos. Env. 31, 2225- 2233.

Da Silveira L., Sternberg L., Mulkey S.S. and Wright S.J. (1989) Oxygen isotope ratio stratification in a tropical moist forest Oecologia 81, 51-56.

Smith F.A. and Freeman K.H. (2006) Influence of physiology and climate on δD of leaf wax n-alkanes from C3 and C4 grasses. Geochim. Cosmochim. Acta 70, 1172- 1187.

Smith F.A., Wing S.L. and Freeman K.H. (2007) Magnitude of the carbon isotope excursion at the Paleocene-Eocene thermal maximum: The role of plant community change. Earth Planet. Sci. Lett. 262, 50-65.

Spicer R.A. (1981) The sorting and deposition of allochthonous plant material in a modern environment at Sillwood Lake, Sillwood Park, Berkshire, England. USGS Surv. Prof. Paper 1143.

Spicer R.A. and Wolfe J.A. (1987) Plant taphonomy of late Holocene deposits in Trinity (Clair Engle) Lake, northern California. Paleobiol. 13, 227-245.

Spicer R.A., Rees P.M. and Chapman J.L. (1993) Cretaceous phytogeography and climate signals. Proc. R. Soc. London, Ser. B 341, 277-285.

Stewart W.N. and Rothwell G.W. (1993) Paleobotany and the Evolution of Plants. Cambridge University Press, Cambridge UK.

Swift M.J., Heal O.W. and Anderson J.M. (1979) Decomposition in Terrestrial Ecosystems. University of California Press, Berkeley, CA.

Tiffney B.H. (1984) Seed size dispersal syndromes, and the rise of the angiosperms: Evidence and hypothesis. Ann. MO Bot. Garden 71, 551-576.

Tipple B.J. and Pagani M. (2010) A 35 Myr terrestrial higher plant n-alkane stable carbon and hydrogen isotope record from the Gulf of Mexico: implications for North

24 American C4 grasslands and hydrologic cycle dynamics. Earth Planet Sci. Lett. 299, 250–262.

Turner I.M. (2001) The Ecology of Trees in the Tropical Rainforest. Cambridge University Press, Cambridge UK.

Upchurch G.R. and Wolfe J.A. (1987) Mid-Cretaceous to early Eocene vegetation and climate: evidence from fossil leaves and woods. In The Origins of Angiosperms and their Biological Consequences (eds. E.M. Friis, W.G. Chaloner and P.R. Crane). Cambridge University Press, Cambridge UK. pp. 75-105.

Valladares F. and Niinemets Ü., (2008) Shade tolerance, a key plant feature of complex nature and consequences. Ann. Rev. Biol. Ecolog. Evolution, and Systematics 39, 237-257.

Vogel J.C., (1978) Recycling of carbon in a forest environment. Oecologia Plantarum 13, 89–94.

Xiao L., Yang H., Sun B., Li X. and Guo J. (2009) Stable isotope compositions of recent and fossil sun/shade leaves and implications for palaeoenvironmental reconstruction. Rev. Palaeobot. Palynol. 190, 75-84. van Aarssen B.G.K., Alexander R. and Kagi R.I. (2000) Higher plant biomarkers reflect palaeovegetation changes during Jurassic times. Geochim. Cosmochim Acta. 64, 1417-1424. van Bergen P.F., Collinson M.E. and Briggs D.E.G. (1995) Resistant biomacromolecules in the fossil record. Acta Bot. Neerlandica 44(4), 319-342.

Wing S.L., Harrington G.J., Smith F.A., Bloch J.I., Boyer D.M. and Freeman K.H. (2005) Transient floral change and rapid global warming at the Paleocene-Eocene boundary. Science 310, 993-996.

Wing S.L., Herrera F., Jaramillo C.A., Gómez-Navarro C., Wilf P. and Labandeira C.C. (2009) Late Paleocene fossils from the Cerrejón Formation, Colombia, are the earliest record of Neotropical rainforest. Proc. Natl. Acad. Sci. 106, 18627-18632.

Wright S.J. (2002) Plant diversity in tropical forests: a review of mechanism of species coexistence. Oecologia. 130, 1-14.

Wright S.J. (2002) Plant diversity in tropical forests: a review of mechanism of species coexistence. Oecologia. 130, 1-14.

Wynn J.G. (2007) Carbon isotope fractionation during decomposition of organic matter in soils and paleosols: Implications for paleoecological interpretations of paleosols. Palaeogeo. Palaeoclimatol. Palaeoecol. 251, 437-448.

25

Zimmerman J.K. and Ehleringer J.R. (1990) Carbon isotope ratios are correlated with irradiance levels in the Panamanian orchid Catasetum viridiflavum. Oecologia 83, 247-249.

26 Chapter 2. Isotopic characteristics of canopies in simulated leaf assemblages

2.1. Abstract

The geologic history of closed-canopy forests is of great interest to paleoecologists and paleoclimatologists alike. Closed canopies have pronounced effects on local, continental and global rainfall and temperature patterns. Although evidence for canopy closure is an elusive feature in the fossil record, the characteristic isotope gradients of the “canopy effect” could be preserved in leaves and proxy biomarkers. To assess this, we employ new carbon isotopic data for leaves collected in diverse light environments within a deciduous, temperate forest (Maryland, USA) and a perennially closed canopy, moist tropical forest (Bosque Protector San Lorenzo, Panamá). In the tropical forest, leaves

13 exhibit a maximum carbon isotope range of 10‰, with higher δ Cleaf values occurring both in upper reaches of the canopy, and with higher light exposure and lower humidity.

Leaf fractionation (∆leaf) co-varied with height above ground, light, and humidity.

13 Vertical C enrichment in leaves largely reflects changes in ∆leaf, and does not trend

13 strongly with δ C of CO2 within the canopy. At the site in Maryland, leaves express a more modest δ13C range (~6‰), with a clear trend that follows both light and elevation.

Using canopy data, binned by elevation, we simulated isotope patterns for leaf assemblages. The re-sampling (bootstrap) model determines both the mean and range of carbon isotope values for simulated leaf assemblages ranging in size from 10 to over

1,000 leaves. Each canopy’s isotope range is captured with 50 or more randomly sampled

27 leaves, thus illustrating that it is possible to distinguish leaf assemblages from a closed canopy with sufficiently large samples. For very large leaf assemblages, mean isotopic values approximate the δ13C of carbon contributed by leaves to soil and are similar to

13 observed δ Clitter values at sites within Panamá, including the canopy crane site. The

13 model predicts a persistent ~1‰ difference in δ Clitter for the two sites consistent with higher water availability in the tropical forests. This work provides a new framework for linking contemporary ecological observations to the geochemical record using flux- weighted isotope data and lends insights to the effect of forest architecture on organic and isotopic records of ancient terrestrial ecosystems.

2.2. Introduction

Dense or “closed” forest canopies are of enormous climatic and ecological significance.

Canopy closure refers to the extent to which the forest floor is covered by vegetation. A common definition of a closed canopy is a minimum coverage of 40% (FAO/UNEP,

1999). Canopy closure affects surface albedo, atmospheric circulation, surface roughness, and hydrologic cycling, which can, in turn, influence terrestrial temperature and rainfall redistribution (Bruenig, 1989; Bastable et al., 1993; van Dijk & Keenan, 2007; Boyce &

Lee, 2010; Boyce et al., 2010). Closed canopies stabilize soils, giving them large above- and below-ground carbon storage potential. By preventing soil erosion and promoting soil moisture, canopy structure can affect soil carbon and the carbon cycle - both on the local and global scales (Grace et al., 1995; Buchmann et al., 1997; Koch, 1998;

Giambelluca, 2002; Reilly et al., 2010). Dense forest canopies also host a great variety

28 of microhabitats into which much of the diversity of animal and plant life has specialized

(Kohyama, 1993; Wilson, 1994; Ozanne et al., 2003), and the proliferation of canopy habitats may have been important in the evolution of arboreal primates and our hominid ancestors (Sussman, 2005).

The three-dimensional structure of a canopy – the height, density, distribution, and coverage of aerial biomass (Parker, 1995) – remains an elusive collection of traits to detect in the fossil record. Fossil assemblages seldom preserve branchless boles, or evidence of tree spacing or branch density associated with canopy density (Secord et al.,

2008). While a few exceptional assemblages include large, fleshy fruits (an adaptation to dense canopy) (Tiffney, 1984; Erikksson et al., 2000; Bruun & Ten Brink, 2008; Friis et al., 2011) or in situ litter, logs, and stumps (Williams et al., 2003; DiMichele & Falcon-

Lang, 2011), the most common macroscopic plant fossils are leaves and leaf fragments

(Ellis et al., 1999). Forest architecture is inferred from leaf fossils by morphologic and taxonomic comparison to extant ecosystems (Upchurch & Wolfe, 1987; Wing et al.,

1991; Wilf, 2000; Wing et al., 2000; Johnson & Ellis, 2002; Wing et al., 2009).

The origin of the closed forest canopy is generally linked with the rise of angiosperms

(Morley, 2000; Burnham & Johnson, 2004) but the emergence and evolution of closed canopies are poorly understood. Indeed, even the timing of these first canopies is disputed (Upchurch & Wolfe, 1987; Davis et al., 2005). It is difficult to determine from the fossil record the character of the canopied forest habitats and how canopies responded to climate perturbations or other events (Boyce et al. 2010). Given the impact of forests

29 on climate systems, there is great interest in identifying canopy properties of ancient ecosystems. A proxy method that captures carbon isotopic gradients could enable characterization of ancient canopy closure (Ehleringer et al.,1986; Secord et al., 2008;

Cerling, et al., 2011). At present, however, the specific imprint of forest architecture on isotopic and geochemical properties of preserved leaf fossils, bulk organic matter or plant biomarkers, is not well known.

A long recognized indicator of canopy conditions - the “canopy effect” - is a decrease in

13 13 δ C of leaves (hereafter called δ Cleaf) from the top of the canopy to the forest floor

(Vogel, 1978). Gradients in light, humidity, and the concentration and carbon isotopic

13 composition of atmospheric CO2 can all influence δ Cleaf (Aoki, 1978; Medina &

Minchin, 1980; Madigosky, 2004; Ometto et al., 2006). Higher carbon assimilation rates in the well-lit canopy top cause leaves to draw down internal CO2 rapidly, thus lowering discrimination against 13C compared with leaves in the understory (Chazdon & Fetcher,

1984; Ehleringer et al., 1986; Zimmerman & Ehleringer, 1990; Hanba et al., 1997;

Lockheart et al., 1997; Heaton, 1999; Niinemets et al., 1999; Meir et al., 2002; Poorter et al., 2006; Niinemets, 2007). This is a major cause of the canopy effect (Baldocchi, 1993;

Ellsworth & Reich, 1993). In addition, lower humidity in the upper canopy stimulates stomatal closure, which slows water loss and restricts the supply of CO2 to the leaf, limiting the extent to which Rubisco (the primary enzyme of carbon fixation in

12 photosynthesis) can express its preference for CO2 (Madhavan et al., 1991; Stewart et al., 1995; Brooks et al., 1997). Finally, understory atmospheric CO2 can be more

13 depleted in C compared to the canopy, due to a higher proportion of respired CO2

30 (Vogel, 1978; Medina & Minchin, 1980; Sternberg et al.,1989; Van der Merwe &

Medina, 1991).

The canopy effect is potentially preserved in the fossil record as a large range in the stable carbon isotope composition of different leaves from the same fossil assemblage

(Arens et al., 2000; Beerling & Royce, 2002; Fricke et al., 2007). The relative abundance and demographics of leaves in living closed canopy forests, however, suggest that the canopy effect might be attenuated and difficult to detect in the fossil record. In tropical closed-canopy forests, the vast majority of leaf biomass occurs in the well-lit upper canopy. Leaves in the upper canopy have higher photosynthetic rates and are short- lived compared to more shaded leaves, although this is not always observed (Parker et al., 1989; Wright & Cannon, 2001; Santiago & Wright, 2007). Reich et al. (1991) found that on a global scale there exists an inverse logarithmic relationship between leaf life span and net photosynthetic rate. Such a production bias may result in a greater proportion of leaves from the upper canopy strata contributing to litter assemblages.

Further, canopy leaves are smaller and exposed to higher winds, making them more likely to be transported to depositional environments where they have a greater likeliehood of entering the fossil record (Dilcher, 1973; Spicer, 1980; Ferguson, 1985;

Burnham, 1992).

This work seeks to understand how taphonomy – the factors interposed between a living ecosystem and a fossil assemblage - affects the isotopic composition of fossil leaf assemblages or sediments. We employ leaf carbon isotope data and published studies of

31 leaf fluxes from two forests with markedly different canopy structure and light environment: 1) a seasonally closed-canopy, mature deciduous temperate forest in coastal

Maryland and 2) an old-growth, perennially closed-canopy, moist evergreen tropical forest in Panamá. We evaluated data for these forests with a re-sampling (bootstrap) model that simulates leaf assemblages in order to answer the following questions.

• How many leaves from a litter assemblage are necessary to distinguish the

isotopic gradient characteristics of canopy closure?

13 • Are mean δ Cleaf values for a litter assemblage diagnostic of canopy type?

• Can we predict the δ13C values of cumulative litter, soil organic matter, and

organic carbon in sedimentary archives using litter flux and isotope patterns in

canopies?

13 We use the re-sampling model to determine the likely δ C range and mean for leaves sampled from a litter assemblage. We also extend our re-sampling model to predict the isotopic composition of cumulative carbon delivered to soils and sediments and compare these results to available data from forest soils. The modeled leaf and soil organic carbon isotope patterns will help improve inferences about forest structure that are derived from carbon isotope measurements of fossil leaves, as well as secondary material – such as teeth, hair, paleosol carbonates, or organic soil carbon (van der Merwe & Medina, 1991,

Koch, 1998; Secord et al., 2008, Levin et al., 2011).

32 The forests chosen for this study occur in very different environments, Panamá and

Maryland, USA. These sites have distinctive differences in climate and seasonality that are reflected in the canopy isotope gradients. These isotopic differences are driven by the timing of leaf production. In the tropical forest, leaves are added throughout the year and canopy closure is persistent (Leigh, 1975; Lowman and Wittman, 1996). In the temperate forest foliage is added within an open canopy during the spring in response to photoperiod and/or temperature (Korner & Basler, 2010).

2.3. Methods

2.3.1. Collection sites

The Bosque Protector San Lorenzo forest preserve of Panamá (Fig. 2.1) is located at 9˚

17’N, 79˚38’W. This evergreen, moist tropical terre firme forest is 130 m above sea level and 4.4 km from the Caribbean coast. The site has a tropical monsoonal climate (Köppen classifcation by FAO GeoNetwork) receiving on average 330 cm of rainfall annually and has a mean annual temperature of 26 ºC. The 6 ha preserve contains more than 22,000 trees and 240 recorded species of trees and lianas. The canopy is co-dominated by

Manilkara spp., and Brosimum utile. Calophyllum longifolium, and Aspidosperma cruenta occupy much of the mid-canopy, and Virola sebifera occupies mid-canopy, understory, and gaps (CTFS; http://www.ctfs.si.edu/site/Sherman). This forest exhibits a marked, year-round light gradient with less than 1% of available light (>99% light

33 attenuation) reaching the forest floor. No logging has occurred on this site for at least 200 years and it has been an officially protected site for over a century.

The Smithsonian Environmental Research Center (SERC) forest (38˚ 53’ N, 76˚ 33’ W) is located on the inner coastal plain of the Chesapeake Bay in Maryland (USA) (Fig. 1).

Mean annual temperature at the site is 13 ºC with hot, humid summers and short, cool winters. Average annual precipitation in this area is 108 cm and the region is classified as a humid subtropical climate (Köppen classifcation by FAO GeoNetwork). This seasonally deciduous forest is dominated by Liriodendron tulipifera in the upper and mid-canopy, Fagus grandifolia in the lower mid-canopy and Carpinus caroliniana in the understory (Parker et al., 1989; Brown & Parker, 1994). This forest canopy is closed

(>99% light attenuation) at the height of the summer growing period but most deciduous foliage is produced during spring leaf flush when the canopy is open (Brown & Parker,

1994). Thus, due to timing, isotopic differences between the temperate and tropical data sets represent fundamental differences in canopy closure. This site has been undisturbed for ~150 years.

2.3.2. Sampling methods and environmental measurements

At the San Lorenzo forest, leaves were collected along a 47-m vertical transect using a canopy crane. This tower crane, managed by the Smithsonian Tropical Research Institute

(STRI), provides access to nearly the entire forest structure in a 0.92 ha area. Bulk leaf samples were taken at 72 unique sampling sites characterized by the average of percent

34 available light during the time of maximum photosynthesis. Leaves were clipped from trees, stored in paper bags and dried at 70ºC the same day as collection. Atmospheric air samples were collected within the canopy in association with leaf sampling sites in 160 ml serum bottles. These bottles were allowed to equilibrate with atmosphere for 60-90 minutes, and then crimp-sealed with butyl rubber gas-tight septa. All samples were acquired during a two-month time window, from January to February of 2010.

Available percentage of light was determined at each sampling site within the canopy from 9 a.m. – 12 p.m. – the portion of the day when the most photosynthesis occurs

(Goulden et al., 2003). Light measurements were made using data from three light meters over a range of time scales. A LiCor LI200SB pyranometer installed on the crane well above the canopy collected mean, maximum, minimum, and total solar radiation at 15 minute time intervals. A second set of light measurements were made using a LiCor

LI190 quantum PAR (photosynthetically active radiation) sensor, recording 5 minute averages, maxima, and minima at each sampling site. Finally, a Spectrum Technologies

Field Scout hand-held quantum PAR meter was used to characterize leaf sites during collection. Light measurements were made repeatedly by all three devices throughout the day and for the entire sampling period in order to characterize fully the average irradiance during the period of highest photosynthesis, before water stress inhibits stomatal opening.

Light data for the tropical forest are summarized in Figures 2A and 2C. Humidity and temperature measurements were coordinated with light measurements using a Traceable digital thermo-hygrometer. Relative humidity values are shown in Figures 3 and 4.

35 Leaf samples from the SERC site in Maryland were collected over a 40-m vertical transect using a mobile construction crane. Leaf samples were taken at 81 unique sampling sites characterized by light environment. Leaves were clipped from trees, stored in paper bags and dried at 70ºC within 8 hours of collection. All samples were acquired during the summer of 2001. Available light at each sampling site was characterized using digital hemispherical photography. PAR is calculated from these images using measurements of cover from direct, indirect, and total site factors, as determined by the

HemiView software package. Site factors reflect the calculated position of the sun relative to obstacles in an image and reflect the mean fraction of all times when light is not blocked during clear (‘direct’) conditions or some standard overcast (‘indirect’) conditions. An average of these two is the ‘total’ site descriptor. Light was binned into five distinct categories of estimated environmental light, the crown illumination index, using methods similar to those outlined in Keeling & Phillips (2007).

2.3.3. Isotopic analysis methods

The carbon isotope analyses of leaves collected in Panamá were performed at

Pennsylvania State University. A composite of dried leaves from each sampling site was ground and homogenized before analysis using a continuous flow Costech elemental analyzer connected to a Thermo Finnigan Delta XP IRMS. δ13C values were corrected for sample size and reported relative to Vienna Pee Dee Belemnite (VPDB) (Coplen et al.,

2006). Error was determined by repeated analysis of internal standards (calibrated by

36 IAEA CH-7, IAEA CH-6, and USGS-24). Instrument precision is ±0.07‰ (n=36).

Accuracy, the average difference between the measured and true δ13C value, is ±0.11‰

(n = 28).

The carbon isotope analyses of leaves collected in Maryland were performed at the

Carnegie Institution of Washington. Dried leaves from each sampling site were ground and homogenized before analysis using a Carlo Erba Instruments NC 2500 elemental analyzer coupled to a Thermo Scientific Delta V Plus IRMS. δ13C values were corrected for sample size and reported relative to VPDB. Instrument precision of ~0.25‰ is based on the reproducibility of NIST standards repeatedly analyzed as samples (NBS-18, NBS-

19, and Iso-Analytic R022).

Carbon dioxide in atmospheric gas samples was cryogenically distilled from water and non-condensible gases under vacuum. Isolated CO2 was isotopically analyzed using a

Finnigan MAT 252 dual-inlet isotope ratio mass spectrometer equipped with a microvolume inlet. Data were corrected for sample size dependence and normalized to the standard Vienna Standard Mean Ocean Water (VSMOW) scale. Isotope values for leaf carbon and CO2 were used to compute Δleaf,

13 13 13 3 ∆leaf = (δ CCO2 - δ Cleaf) / (1+ δ Cleaf /10 ) (Eqn. 1)

the fractionation during carbon fixation due to gas diffusion at the stomata and enzymatic fractionation by Rubisco (Farquhar et al., 1989). The Δleaf notation for fractionation is

37 common in ecological literature and is a close approximation of the negative of εleaf:

$ δleaf +1000 ' εleaf = 1000& −1) (Eqn. 2) %δ CO2 +1000 (

€ whereas εleaf values are more commonly used in geochemical literature.

2.3.4. Flux and biomass

Vertical biomass distribution data are not available for the San Lorenzo site in Panamá.

We therefore employed data from a similar forest (Osada et al., 2001) to estimate standing biomass and annual leaf flux. The Pasoh Forest Reserve at 2˚ 59’ N, 102˚ 18’ E in Peninsular Malaysia is a well-studied site that is widely used to infer and compare properties of lowland tropical rainforests (Manokaran & LaFrankie, 1990; Condit et al.,

1996; Condit et al., 1999; Wills & Condit, 1999; Wright et al., 2005). Pasoh is a moist evergreen tropical forest and the canopy has a similar stand structure to the Panamanian forest (Manokaran & LaFrankie, 1990). The Pasoh forest community is highly diverse and dominated by Dipterocarpaceae. Shorea spp. comprise most of the upper canopy with Xerospermum noronhianeae and Dacryodes spp. prominent in the mid-canopy.

Pasoh biomass and annual leaf flux data were originally published by Osada et al. (2001).

Biomass refers to the total mass (tons) of leaf material per area (hectare) on the trees at the height of the growing season and flux is defined as the annual mass of leaves per area from within a height interval (1 m) that falls as litter. The authors estimated forest

38 biomass using equations that relate leaf biomass to diameter and height of bole. These equations were derived from destructive sampling at a small site and extrapolated to the rest of the stand. Leaf fall flux was estimated by relating leaf life span at a particular height to the biomass at that height (Osada et al., 2001).

While the two tropical closed-canopy forests used for biomass and flux data (Pasoh

Forest; Osada et al., 2001) and for isotopic data (San Lorenzo; this study) are populated by different taxa, the stand structures are very similar (Manokaran & LaFrankie, 1990;

Condit et al, 1996.). The Osada et al. (2001) data set for biomass and flux at Pasoh is of sufficient vertical resolution to be compared to our data from Panamá and from the temperate Maryland forest at SERC (Parker et al., 1989). Biomass and flux data for the

SERC forest were originally published by Parker et al. (1989); leaf sampling and isotopic analyses are more recent.

2.3.5. Bootstrap analysis

13 Leaf flux and δ Cleaf values were incorporated into a resampling model using the ‘R’ statistical package (R Core Development Team, 2007). All data were binned in five meter increments vertically, resulting in nine bins for the San Lorenzo forest and seven for the

SERC forest. The resampling model performs a bootstrap analysis of binned flux and isotope data for both forests.

The resampling model samples with replacement biomass flux and isotopic data binned

39 by height. The resampling model chooses a virtual ‘leaf’ from these bins with the likelihood (L) of selecting a leaf from a particular vertical bin (bin of height x) determined by the proportion of leaf flux coming from that bin.

L = Φbin x / Φ∑ all bins (Eqn.3)

The collected ‘leaf’ is then assigned a randomly selected δ13C value from the collected data for that height bin in that forest type. The model repeats sampling and assignment for y number of times, where y is the assemblage or collection size – the number of

‘leaves’ sampled in silico. The mean δ13C value for a single assemblage of y leaves

13 (δ Cy-mean) is then calculated and recorded. Each simulated collection scenario is

13 repeated 2000 times resulting in 2000 assemblage δ Cy-mean values. The highest 50 and

13 lowest 50 of these δ Cy-mean values are removed from the data set to include only those means within 95% of the collected data, approximately two standard deviations. The isotopic range of leaves (Δy) for a given sampling size y is calculated as the difference

13 13 13 between the highest and lowest δ Cleaf values (δ Cy-max mean - δ Cy-min mean = Δy) from the remaining 1900 values.

2.4. Results

2.4.1. Light and height measurements

40 Figures 2.2A and 2.2B summarize vertical light data for both forest types, and both illustrate the general trend of increasing light with increasing height. Further, Fig. 2.2A also shows the very wide diversity of light environments available to leaves in the dense mid-canopy of a tropical closed-canopy forest. Similarly, the temperate forest (Fig. 2.2B) is characterized by three distinct light regions (0-15 m, 15-25 m, and 25-35 m) dictated by the bimodal biomass distribution and a mid-canopy with a variety of levels of available light.

2.4.2. Leaf carbon isotope data

13 δ Cleaf measurements from Maryland forest (Fig. 2.2D) are generally more enriched

(~1‰) than leaves sampled from the Panamá forest (Fig. 2.2C). Both forests exhibit a

13 vertical isotope gradient (i.e., the canopy effect): δ Cleaf values are lowest near the ground and trend to higher values upward. The extent and character of this gradient differs between the sites. The Maryland forest has an overall range of 5.8‰, the Panamá

13 forest an overall range of ~ -10‰, but the Panamá site has much more variable δ Cleaf values (Fig. 2.2D). The most 13C-depleted leaves (~ -34‰) are observed in understory samples collected less than 5 m above the forest floor. The high isotopic variability as a function of height likely reflects the highly variable growth conditions within the upper and mid-canopy (i.e., 20 to 30 m above the forest floor), where light ranges between 10

13 and 90% and the δ Cleaf values range nearly 6‰.

41 2.5. Discussion

2.5.1. Light properties and the δ13C canopy effect at San Lorenzo

13 The canopy effect is commonly attributed to C-depleted CO2 that collects at the base of a forest as a result of microbial respiration in the soil (Vogel, 1978; Medina & Minchin,

1980; Van der Merwe & Medina, 1991). This observation is supported by atmospheric

13 sampling conducted in many forests that shows overall depletion in C of CO2 during the night as photosynthesis ceases but microbial respiration continues (see Pataki et al.,

2003). Although such data are not available for our study, we sampled atmospheric gases in the mid-canopy of the San Lorenzo forest in order to assess the variation of

13 δ CCO2 with height, humidity, and light (as a measure of canopy density and atmospheric mixing) at the time when photosynthetic activity is highest (before 12 pm; Goulden et al.,

2003) (Fig. 2.3A, C, E). Despite an expectation of a positive relationship between light

13 and the δ CCO2, our data show no significant correlation between these two variables, or

13 between height and δ CCO2 (Fig. 2.3A, C). This suggests that the atmosphere was well mixed at mid-day.

13 Using our CO2 data, we determined ∆leaf using δ Cleaf values for a leaves at the atmosphere collection site (this collection subset includes Aspidosperma, Brosimum,

Calophyllum and Virola). ∆leaf is plotted in figures 2.3B and 2.3D relative to leaf sample height and available light. ∆leaf, as defined by Farquhar et al., (1989) (Eqn. 1) describes the fractionation of carbon isotopes by a leaf during the uptake and fixation of CO2.

42 Greater ∆leaf values are expected in leaves with higher stomatal conductance, such as those in humid forest settings, and those with low rates of photosynthesis and carbon fixation because of low light levels. In both situations CO2 within the leaf is replenished rapidly relative to the rate at which it is fixed, allowing Rubisco to more fully express its

12 preference for CO2. As expected, our data reveal greater fractionation (i.e., larger ∆leaf values) at both lower light availability and with lower leaf height (Fig. 2.3B, D).

Reduced isotope fractionation with height contributes to the observed canopy effect. As expected, higher relative humidity correlates with increased fractionation (larger ∆leaf values, Fig. 2.3F) since higher humidity encourages greater stomatal opening and allows for greater expression of enzymatic fractionation by Rubisco (Madhavan et al., 1991).

13 Correlation between light and δ CCO2 values likely reflects the influence of photosynthetic uptake rate (Fig. 2.3E). The relative effects of humidity and light on photosynthesic fractionation is difficult to gauge within this small section of the canopy, since both are humidity and light are highly variable and weakly co-vary within the dense canopy (Fig. 2.4).

A stepwise regression model of light and relative humidity indicates that light is more

2 = strongly associated with the ∆leaf trend (R 0.40; p =0.02). Adding the relative humidity data only marginally improved the predictive ability of this model (R2 = 0.41; (with light alone, P=0.06). These results indicate that light level at the leaf has more effect on fractionation than the relative humidity at that site. The results of other stepwise

13 regression models for δ CCO2 that considered factors such as respiration or carbon

43 assimilation rates were not statistically significant indicating these factors are not

13 influential in the δ CCO2 values of the intra-canopy atmosphere.

13 Given that both height and light trend with ∆leaf, we suggest that lower δ C leaf in the lower canopy reflects slower photosynthetic rates as well as higher humidity and lower vapor pressure deficits at depth in the lower canopy.

2.5.2. Canopy isotope gradients in biomass and litter flux

Figure 5 illustrates biomass allocation, annual flux and carbon isotope composition for both temperate and tropical forests. The Maryland forest exhibits a bimodal biomass distribution (Fig. 2.5A; Parker et al., 1989) with thick understory and upper canopy layers. By comparison, tropical forests (Osada et al., 2001) (Fig. 2.5D) have most of their biomass in the upper canopy and very little is located in the understory (Fig. 2.3D;

Osada et al., 2001). Consistent with other studies (Lowman and Wittman, 1996; Leigh,

1999) low-latitude closed-canopy forests (Fig. 2.5E) have a much larger foliar biomass flux when compared to high latitude closed-canopy forests. For the temperate forest, annual biomass flux (382.66 g/m2; Fig. 2.3B) differs only slightly from the standing biomass estimate (Fig. 2.5A) because the forest trees are annually deciduous and herbivory is low at this site (Parker et al., 1989). Biomass flux in the tropical forest (829 g/m2; Fig. 2.5E) illustrates the shorter lifespan and rapid turnover of leaves in the upper canopy associated with high irradiance and very high rates of photosynthesis.

44 Leaves in the mid-canopy (20-30 m) of the tropical forest exhibit a wide array of δ13C

13 values (Fig. 2.5F) whereas δ C leaf values from the Maryland forest (Fig. 2.5C) exhibit a more linear relationship with light. The Maryland forest has less variability in carbon isotopic composition at any given elevation. This may be the influence of enhanced atmospheric mixing or the result of more even lighting throughout the canopy during the time of spring leaf flush (Polgar & Primack, 2011).

2.5.3. Resampling model outcomes

How many leaves from a litter assemblage are necessary to distinguish the isotopic gradient characteristics of canopy closure?

The understory produces a low flux of leaves to depositional environments. As a result, a relatively large number of leaves are likely necessary to find even one with the lowest carbon isotope composition characteristic of the canopy isotope gradient. Thus, our aim was to quantify the minimum sample size needed to ensure a high probability of both including leaves from the understory and capturing the isotopic range diagnostic of canopy closure. We simulated assemblages of y = 10, 25, 50, 100, and 250 leaves - amounts most likely to be available in paleoflora (compiled in Wing et al., 2009),

The isotopic range of leaves within an assemblage was calculated as the difference

13 between the highest and lowest δ Cleaf values (Δy) for the 95% of the leaf population returned by the computational model. Although mean values for Δy consistently differ by

45 ~1‰ between forest types (Fig. 2.6), Δy values for collections of y ≤ 50 overlap considerably. For larger sample sizes (y >50), mean Δy values for simulated collections become more distinct between the two forest types until there is no overlap in Δy.

Overall, our modeling results indicates that it is possible to capture the full range of the canopy effect from a random sampling of y > 50 leaves. The percentage of simulated collection assemblages from the tropical data set that included at least one understory leaf

(i.e., those leaves from the lowest bin) was found to be 14.9% for y = 25, 28.8% for y =

50, 49.9% for y = 100 and 84.6% y ≥ 250. A samples size of y > 50 leaves is manageable for many paleofloral assemblages and 28.8% of simulated assemblages included an understory leaf. Further, the odds of finding understory leaves increase when leaf selection is based on morphologic metrics of light environment (e.g., size, vein density stomatal index, etc.) of whole leaves from well-described families (Upchurch & Wolfe,

1987; Kubiske et al., 1996; Richards, 1996; Boyce, 2008; Dillen et al., 2008).

13 Are mean δ Cleaf values for a litter assemblage diagnostic of canopy type?

Within a canopy type, the mean carbon isotope composition of simulated assemblages

13 (δ Cy-mean) are similar and reflect the dominance of upper canopy biomass in all leaf collections where y > 50 (Fig. 2.7).

13 For leaf assemblages of all sizes (y = 10 to 1,000), δ Cy-mean values for the tropical forest are consistently 1‰ lower than for the temperate forest (Fig. 2.7). This observation is not

46 13 13 due merely the input of the highly C-depleted, shade leaves. For example, δ Cy-mean values were similarly 1‰ depleted even for very small sample sizes (y = 10) where only

5.5% of collection events included at least one understory leaf. Even large assemblages contain very few understory leaves, and for y = 1,000, assemblages contain no more than

5.1% percent understory leaves.

The canopy between 20-30 m contributes 55% of leaf flux (Osada et al., 2001) and dominates the carbon isotope composition of simulated leaf assemblages. This region

(Fig. 2.2A) experiences highly variable conditions that range from open gaps to highly

13 dense foliage (Brokaw, 1982).The variable conditions yielded a wide range of δ Cleaf values (Fig. 2.2C) which have a mean value of -30‰ and a calculated standard deviation

13 approaching 4‰. Further, observed δ Cleaf values for the uppermost canopy in the tropical forest are 1-2 ‰ lower than the leaves from the very top of the temperate forest

(Fig. 2.2C and 2.2D). This is consistent with high humidity and greater fractionation

(∆leaf) in tropical rainforests (Diefendorf et al., 2010). Leaves in the mid canopy at the

Panamá site have ∆leaf values that range from 18 to 26‰, consistent with low light and overall high, mid-canopy humidity and (Fig. 2.4; Diefendorf et al., 2010). The small

13 range of δ Cleaf values and overall lower Δleaf values in the Maryland forest may be due to the greater air mixing or lower humidity in the temperate canopy (Stewart et al., 1995;

Diefendorf et al., 2010).

Can we predict the δ13C values of cumulative litter, soil organic matter, and organic carbon in sedimentary archives using litter flux and isotope patterns in canopies?

47

A simulated collection of many thousands of leaves estimates the integrated δ13C signal leaf litter imparts to soil and sedimentary organic carbon stores. High numbers of leaves

13 13 (y ≥ 1,000) converge at distinct δ Cy-mean values. The δ C composition of litter is characteristic of the forest site and represents the combined influenced of climate, vegetation, and the architectural differences between the forest (Diefendorf et al., 2010).

13 Santiago (2003) reported δ Clitter values ranging from -30.5 to -28.4‰ for 16 samples

13 from four locations in and near San Lorenzo preserve; the author also reported δ Csoil

(A-horizon) values that ranged from -27.5 to -26.3‰ for 24 samples from four locations in and near the preserve. Measured carbon isotope values for litter collected in

Panamanian forests (-29.7 ± 1‰) are indistinguishable from the predicted value -29.3 ±

13 13 0.3‰ based on our δ Cleaf data (Fig. 8). δ Clitter differ considerably from the values

13 measured in δ Csoil.

13 13 13 In a compilation of published δ Cleaf, δ Clitter and δ Csoil data, Bowling et al. (2008)

13 13 found δ Clitter values to differ from bulk δ Cleaf values by -1.5‰ to +3‰. This range likely indicates the degree of decomposition in the litter collection and could be biased by the limited number of leaves in some litter samplings (Nadelhoffer & Fry, 1988). By

13 13 extension, they found δ Csoil was up to 4.5‰ more enriched than δ Cleaf. The isotopic difference between litter and soil carbon in tropical forests determined by Santiago

(2003) was +2.8‰ - within the range found by Bowling et al. (2008). Bowling et al.

(2008) included only sun leaves, which are generally more enriched in 13C (Ehleringer et

48 al., 1986; Figure 2 of this study). While sun-leaves are a large component of leaf flux

(Fig. 2.5E), the upper 10m of canopy accounts for slightly less than half (~45%) of total biomass flux. Thus the dense mid-canopy must be considered in any estimation of soil

13 carbon arising from leaf material. Our bootstrap analysis gives estimates of δ Clitter from a flux-weighted integration of leaves from the entire height of a forest and provides a

13 more accurate estimation of the δ C contribution of leaves to soils. In order to understand the offset between litter and mineral soil (A-horizon) organic carbon it is essential to account for all carbon inputs. Soils are an integrated product of all carbon inputs as well as their products of their degradation.

Santiago (2003) reported isotopic data for litter and soils collected in Panamá from sites with a wide range in annual precipitation (1800-3500 mm/yr). We would expect leaves from drier areas to reflect greater water-use efficiency by having lower foliar carbon isotope compositions (Nadelhoffer & Fry, 1988; Diefendorf et al., 2010). We also note that Santiago’s (2003) litter sampling likely had fewer than 500 leaves, and certainly

13 many fewer than the thousands of leaves our model uses to generate an expected δ Clitter.

Further, Santiago (2003) used a comprehensive scheme of litter collection that included woody debris, frass, and other floral and reproductive organs not included in the modeled

13 litter value. The wide range of measured δ Clitter values potentially reflects contributions from plant components with different chemical and isotopic properties than leaves. In general, across forest types from the tropical to the temperate ecosystems, leaves account for 50-60% of litterfall (Odum, 1970; Jordan & Uhl, 1978; Parker et al., 1989; Leigh,

49 1999; Santiago, 2003). The remaining litterfall comes largely from wood, a material that is ~3‰ enriched with respect to leaf material (Gröcke, 2002).

Most considerations of soil organic carbon fail to recognize the contribution of carbon from roots. Some authors consider root carbon contribution to be greater than or equal to leaves (Fahey et al., 2005; Rasse et al., 2005). Further study is necessary to compare the relative inputs of root, leaf, and biomass carbon as well as the preservation mechanisms for these different carbon pools (Ehleringer et al., 2000; Mueller et al., 2012).

2.5.4. From leaves to soil organic carbon

Soil organic matter is a complex mixture from many inputs and is subject to significant chemical change during decomposition. Plant biochemical components (e.g. leaf lipids, lignin, etc.) are preserved on geological timescales but are also subject to the influence of flux and isotopic gradients associated with canopy coverage. Canopy structure in forests generates a range of light and humidity conditions that are reflected in the isotopic composition of leaves, which transmit this heterogeneity to soils when they are abscised.

Our approach, which considers the biomass flux as well as isotopic composition, offers a framework for understanding the influence of canopy on leaf molecular constituents, provided their abundances are characterized by canopy position. For example, lipid abundance distribution as well as isotopic fractionation differ between sun and shade leaves (Collister et al., 1994; Lockheart et al., 1997; Diefendorf et al, 2011)

50 The modeling approach outlined in this study can be applied to other forests to predict the isotopic composition of the carbon in soils that arise from foliar material. Leaf isotopic compositions, is also affected by the taxa (see Figures 2.2C and 2.2D), regional precipitation regimes (Diefendorf et al., 2010), and canopy coverage (Ehleringer et al.,

2000; Cerling et al., 2011). Spatial changes in moisture, plant composition and canopy

(which generally co-vary) can be captured in soil organic carbon (SOC) at the level of landscape patches (ie. paleoecotones; Nadelhoffer & Fry, 1988; Mariotti & Peterschmitt,

1993; McClaran & McPherson, 1995; Billings, 2006, Cerling et al., 2011, Magill et al.,

2012).

Diagenetic enrichment of 13C in soil organic carbon is rapid (sub-decadal: Bird et al.,

1996) and the temporal resolution of the isotopic composition is driven by the amount of organic input as well as the mobility of the organic matter (McClaran & McPherson,

1995; Boutton et al., 1997; see Wynn, 2007). Our understanding of soil carbon diagenesis is confounded by the many environmental factors that affect this change and hampered by an incomplete understanding of initial organic inputs. This computational approach can improve diagenetic fractionation estimates by providing a better estimate of the initial organic matter within the context of biotic (plant type) and physical (moisture and canopy structure) factors that drive foliar carbon isotope composition.

2.6. Conclusions

51 13 13 We have produced a unique dataset including δ Cleaf, δ CCO2, humidity, height and light for leaves of five major genera in a both a tropical closed-canopy forest and a temperate deciduous seasonal closed-canopy forest. By combining these data with biomass flux measurements for an analogous closed canopy forest ,we compared the isotopic expression of the perennially closed-canopy forest with a similarly sampled seasonally

13 closed-canopy forest. We have found the two forests have a different range of δ Cleaf particular to each canopy type (Maryland forest = ~6‰; Panamá forest = ~10‰) and that it would be possible to detect this diagnostic range with as few as ≥50 randomly sampled leaves from a paleoflora. For very large virtual samplings from both forests, a distinct 1‰ difference in foliar 13C content between forests types is evident, similar to differences observed in litterfall studies and consistent with differences found in biomes with differing water availability (Diefendorf et al., 2010). Foliar material can account for over half the carbon flux to soils and has an important influence on the isotopic character of soil organic carbon. Soil organic carbon can be preserved on very long time-scales, but its δ13C values are subject to diagenetic influences that are not fully understood.

Modeling efforts, such as this study, can aid our understanding of these effects by providing constraints on the δ13C values of leaf litter in the context of forest canopy structure and the resulting differences in environmental drivers such as light, humidity, and source atmospheric CO2.

52 2.7. Figures

Figure 2.1. Two field sites for vertical sampling of leaf material and light measurement.

53

Figure 2.2. Light measurements and foliar δ13C for both tropical and temperate canopy leaf sampling sites. Light at the Panamá site (2A) light is expressed as a percentage of available light whereas light at the Maryland site (2B) is binned in order of increasing irradiance as determined by hemispheric photography. Tropical canopy foliage (2C) spans a ~10‰ range in δ13C that correlates with percentage of available light. δ13C of temperate canopy leaves (2D) also trend with light but express only a ~6‰ range.

54

13 Figure 2.3. Seasonal canopy standing biomass (3A), annual leaf flux (3B), and δ Cleaf data (3C) for 5m vertical bins from Maryland site. Tropical canopy biomass (3D) and annual leaf flux data (3E) are from Pasoh Reserve in Malaysia (Osada et al. 2001). 13 Tropical canopy δ Cleaf (3F) for Panamá site. Isotope ranges are expressed as box-and- whisker plots representing the mean as well as lower and upper quartiles.

55

13 Figure 2.4. δ Cair of isolated atmospheric CO2 from the Panamá site as related to height within the canopy (4A), % light availability (4B), and relative humidity (4C). Carbon 13 13 isotope discrimination (∆leaf ) values is calculated from δ Cair and δ Cleaf measured at each sampling site and related to canopy height (4D), % light availability (4E), and relative humidity (4F).

56

Figure 2.5. Relative humidity as related to height within the Panamá site canopy (5A) and % light availability (as a measure of canopy openness) (5B). Relative humidity is highest in understory but highly variable throughout the mid and upper canopy.

57

Figure 2.6. Absolute isotopic range (∆y) returned by the resampling model for the discrete sample size (y) specified. Box plots show the median, maximum, minimum, and upper and lower quartiles generated by the model. As sample sizes increases expressed range reaches an effective plateau for larger sample sizes and the mean changes very little.

58

13 Figure 2.7. Curves bound the range of foliar δ Cy-mean returned by model for both 13 canopy types. δ Cy-mean for small sample sizes (y) can vary widely in response to rare isotope values but large sample sizes approach a mean that represents the more probable foliar carbon isotope input for leaves in litter assemblages.

59

13 Figure 2.8. Predicted δ Clitter returned by resampling as well as for four litterfall collection sites and six soil samples from a transect of Panamanian forest that includes the study site from this study. Values depicted are the mean and standard error for computational and observational measurements.

60 2.8. References

Aoki M., Yabuki K. and Kohyama H. (1978) Micro-meteorology of Pasoh forest. Malayan Nature Journal 30, 149-159.

Arens N.C., Jahren A.H. and Amundson R. (2000) Can C3 plants faithfully record the carbon isotopic composition of atmospheric carbon dioxide? Paleobiol. 26, 137- 164.

Baldocchi D.D. (1994). A comparative study of mass and energy exchange over a closed (wheat) and an open (corn) canopy: I. the partitioning of available energy into latent and sensible heat exchange. Agricultural and Forest Meteorol. 67, 191-220.

Bastable H.G., Shuttleworth J., Dallarosa R.L.G., Fisch G. and Nobre C.A. (1993) Observations of climate, albedo, and surface radiation over cleared and undisturbed Amazonian forest. Int. J.Climatol. 13, 783-796.

Beerling D.J. and Royer D.L. (2002) Fossil plants as indicators of the Phanerozoic global carbon cycle. Annu. Rev. of Earth Planetary Sci. 30, 527-556.

Bilings S.A. (2006) Changes in stable isotopic signatures of soil nitrogen and carbon during forty years of forest development. Oecologia 148, 325-333.

Bird M.I., Summons R.E., Gagan M.K., Roksandic Z., Dowling L., Head J., Keith Fifield L., Cresswell R.G. and Johnson D.P. (1995) Terrestrial vegetation change inferred from n-alkane δ13C analysis in the marine environment. Geochim. Cosmochim. Acta 59, 2853-2857.

Boutton T.W., Archer S.R. and Midwood A.J. (1999) Stable isotopes in ecosystem science: Structure, function, and dynamics of a subtropical savanna. Rapid Commun. in Mass Spectrom. 13, 1263-1277.

Bowling D.R., Pataki D.E. and Randerson J.T. (2008) Carbon isotopes in terrestrial ecosystem pools and CO2 fluxes. New Phytologist. 178, 24-40.

Boyce C.K. (2008) The fossil record of plant physiology and development - what leaves can tell us. Paleont. Soc. Papers 14, 133-146.

Boyce C.K., Lee J-E., Field T.S., Brodribb, T.J. and Zwieniecki M.A. (2010) Angiosperms helped put the rain in the rainforests: the impact of plant physiological evolution on tropical biodiversity. Ann. MO Botanical Garden 97, 527-540.

61 Boyce C.K. and Lee J-E. (2010) An exceptional role for flowering plant physiology in the expansion of tropical rainforests and biodiversity. Proc. R. Soc. London, Ser. B 277, 3437-3443.

Brooks J.R., Flanagan L.B., Varney G.T. and Ehleringer J.R. (1997) Vertical gradients in photosynthetic gas exchange characteristics and refixation of respired CO2 within boreal forest canopies. Tree Physiol. 17, 1-12.

Brown M.J and Parker G.G. (1994) Canopy light transmittance in a chronosequence of mixed-species deciduous forests. Can. J. Forestry Res. 24, 1694-1703.

Bruenig E.F. (1989) Ecosystems of the World. In: Ecotoxicology and Climate (eds. P. Bourdeau, J.A. Haines, W. Klein and C.R.K. Murti). John Wiley & Sons, Chichester UK. pp. 29–40.

Bruun, H.H. and Ten Brink D.J. (2008) Recruitment advantage of large seeds is greater in shaded habitats. Ecosci. 15, 498-507.

Buchmann N., Kao W-Y. and Ehleringer J. (1997) Influence of stand structure on carbon- 13 of vegetation, soils, and canopy air within deciduous and evergreen forest in Utah, United States. Oecologia 110 109-119.

Burnham R.J., Wing S.L. and Parker G.G. (1992) The reflection of deciduous forest communities in leaf litter: Implications for autochthonous litter assemblages from the fossil record. Paleobiol. 18, 30-49.

Burnham R.J. and Johnson K.R. (2004) South American palaeobotany and the origins of neotropical rainforests. Proc. R. Soc. London, Ser. B 359, 1595-1610.

Cerling T.E., Wynn J.G., Andaje S.A., Bird M.I., Kimutai Korir D., Levin N.E., Mace W., Macharia A.N., Quade J. and Remien C.H. (2011) Woody cover and hominin environments in the past 6 million years. Nature 476, 51-56.

Chazdon R.L. and Fetcher N. (1984) Photosynthetic light environments in a lowland tropical rainforest in Costa Rica. J. Ecol. 72, 553-564.

Collister J.W., Rieley G., Stern B., Eglinton G., and Fry B. (1994) Compound-specific δ13C analyses of leaf lipids from plants with differing carbon dioxide metabolisms. Org. Geochem. 21, 619-627.

Condit R.G., Hubbell S.P., LaFrankie J.V., Sukumar R., Manokaran N., Foster R.B. and Ashton P.S. (1996) Species-area and species-individual relationships for tropical trees – a comparison of three 50-ha plots. J. Ecol. 84, 549-562.

Condit R.G., Ashton P.S., Manokaran N., LaFrankie J.V., Hubbell S.P. and Foster R.B. (1999) Dynamics of the forest communities at Pasoh and Barro Colorado:

62 Comparing two 50 ha-plots. Philos. Trans. R. Soc. London, Ser. B 354, 1739- 1748.

Coplen T.B., Brand W.A., Gehre M., Groning M., Meijer H.A.J., Toman B. and Verkouteren R.M. (2006) New guidelines for δ13C measurements. Anal. Chem. 78, 2439-2441.

Davis C.C., Webb C.O., Wurdack K.J., Jaramillo C.A. and Donoghue M.J. (2005) Explosive radiation of Malpighiales supports a mid-Cretaceous origin of modern tropical rainforests. Am. Naturalist 165, E36-E65.

Diefendorf A.F., Mueller K.E., Wing S.L., Koch P.L. and Freeman K.H. (2010) Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proc. Natl. Acad. Sci. U.S.A. 107, 5738-5743.

Diefendorf A.F., Freeman K.H., Wing S.L., and Graham H.V. (2011) Production of n- alkyl lipids in living plants and implications for the geologic past. Geochim. Cosmochim. Acta 75, 7472-7485.

Dilcher D.L. (1973) A paleoclimatic interpretation of the Eocene floras of Southeastern North America. In Vegetation and Vegetational History of Northern Latin America (ed. A. Graham). Elsevier Pub. Co., Amsterdam. pp. 39-59.

Dillen S.Y. Marron N., Koch B. and Ceulemans R. (2008) Genetic variation of stomatal traits and carbon isotope discrimination in two hybrid poplar families (Populus deltoides ‘S9-2✕ P. nigra ‘Ghoy’ and P. deltoides ‘S9-2✕ P. trichocarpa ‘V24’) Ann. Botany 102, 399-407.

Dimichele W.A. and Falcon-Lang H.J. (2011) Pennsylvanian ‘fossil forests’ in growth position (T-0 assemblages); origin, taphonomic bias and palaeoecological insights. J. Geol. Soc. 168, 585-605.

Ehleringer J.R., Field C.B., Lin Z. and Kuo C. (1986) Leaf carbon isotope and mineral composition in subtropical plants along irradiance cline. Oecologia 70, 520-526.

Ehleringer J.R., Buchmann N. and Flanagan L.B. (2000) Carbon isotope ratios in belowground carbon cycle processes. Ecolog. App. 10, 412-422.

Ellis B., Johnson K.R. and Dunn R.E. (1999) Evidence for an in situ early Paleocene rainforest from Castle Rock, Colorado. Rocky Mountain Geol. 38, 73-100.

Ellsworth D.S. and Reich P.B. (1993) Canopy structure and vertical patterns of photosynthesis and related leaf traits in a deciduous forest. Oecologia 96, 169- 178.

63 Eriksson O., Friis E.M. and Löfgren P. (2000) Seed size, fruit size, and dispersal systems in angiosperms from the early Cretaceous to the late Tertiary. Am. Naturalist 156, 47-58.

Fahey T., Siccama T., Driscoll C., Likens G., Campbell J., Johnson C., Battles J., Aber J., Cole J., Fisk M., Groffman P., Hamburg S., Holmes R., Schwarz P. and Yanai R. (2005) The biogeochemistry of carbon at Hubbard Brook. Biogeochem. 75, 109- 176.

Farquhar G.D., Ehleringer J.R. and Hubick K.T. (1989) Carbon isotope discrimination and photosynthesis. Annu. Rev. Physiolog. Plant Molecular Biol. 40, 503-537.

Fricke H., Ellis B., Johnson K.R. and Jahren A.H. (2007) Carbon isotope characterization of the closed canopy (?) Castle Rock rainforest. Geolog. Soc. of Am. Annu. Meeting, Denver, CO. #6-10(abstr.).

Friis E.M., Crane P.R. and Pedersen K.R. (2011) Early Flowers and Angiosperm Evolution. Cambridge University Press, Cambridge UK.

Giambelluca T.W. (2002) Hydrology of altered tropical forest. Hydrolog. Processes 16, 1665-1669.

Goulden M.L., Miller S.D., da Rocha H.R., Menton M.C., de Freitas H.C., de Silva Figueira A.M. and Dias de Sousa C.A. (2004) Diel and seasonal patterns of tropical forest CO2 exchange. Ecolog. App. 14, S42-S54.

Grace J., Lloyd J., McIntyre J., Miranda A.C., Meir P., Miranda H.S., Nobre C.A., Moncrieff J., Massheder J., Yadvinder M., Wright I. and Gash J. (1995) Carbon dioxide uptake by an undisturbed tropical rainforest in southwest Amazonia, 1992 to 1993. Science 11, 778-780.

Gröcke D.R. (2002) The carbon isotope composition of ancient CO2 based on higher- plant organic matter. Philos. Trans. R. Soc. London, Ser. A 360, 633-658.

Hanba Y.T.S., Lei T.T., Koike T. and Wada E. (1997) Variations in leaf δ13C along a vertical profile of irradiance in a temperate Japanese forest. Oecologia 110, 253- 261.

13 12 Heaton T.H.E. (1999) Spatial, species, and temporal variations in the C/ C ratios of C3 plants: implications for paleodiet studies. J. Archaeolog. Sci. 26, 637-649.

Johnson K.R. and Ellis B. (2002) A tropical rainforest in Colorado 1.4 million years after the Cretaceous-Tertiary boundary. Science 296, 2379-2383.

Jordan C.F. and Uhl C. (1978) Biomass of a terra firme forest of the Amazon Basin. Ecologia Plantarum 13, 387-400.

64

Keeling H.C. and Phillips O.L. (2007) A calibration method for the crown illumination index for assessing forest light environments. Forest Ecol. Manag. 242, 431-437.

Koch P.L. (1998) Isotopic reconstruction of past continental environments. Annu. Rev. Earth Planetary Sci. 26, 573-613.

Kohyama T. (1993) Size-structured tree populations in gap-dynamic forest; the forest architecture hypothesis for the stable coexistence of species. J. Ecol. 81, 131-143.

Korner C. and Basler D. (2010) Phenology under global warming. Science 327, 1461- 1462.

Kubiske M.E., Abrams M.D. and Mostoller S.A. (1996) Stomatal and nonstomatal limitations of photosynthesis in relation to the drought and shade tolerance of tree species in open and understory environments. Trees: Structure and Function 11, 76-82.

Leigh E.G. (1975) Structure and climate in tropical rainforest. Annu. Rev. Ecol. Systematics 6, 67-86.

Leigh E.G. (1999) Tropical Forest Ecology: A View from Barro Colorado Island. Oxford University Press, New York.

Levin N.E., Brown F.H., Behrensmeyer A.K., Bobe R. and Cerling T.E. (2011) Paleosol carbonates from the Omo Group: isotopic records of local and regional environmental change in East Africa. Palaeogeography, Palaeoclimatol., Palaeocol. 307, 75-89.

Lockheart M.J., Van Bergen P.F. and Evershed R.P. (1997) Variations in the stable carbon isotope compositions of individual lipids from the leaves of modern angiosperms: implications for the study of higher land plant-derived sedimentary organic matter. Org. Geochem. 26, 137-153.

Lowman M.D. and Wittman P.K. (1996) Forest canopies: methods, hypotheses and future directions. Annu. Rev. Ecol. Systematic 27, 55-81.

Madhavan S., Treichel I. and O’Leary M.H. (1991) Effect of relative humidity on carbon isotope fractionation in plants. Botany Acta 104, 292-294.

Madigosky S.R. (2004) Tropical Microclimate Consideration. In: Forest Canopies (eds. M.D. Lowman and H.B. Rinker) Elsevier Academic Press, Burlington MA. pp. 24–48.

Magill C.R., Ashley G.M. and Freeman K.H. (2013) Water, plants, and early human habitats in eastern Africa Proc. Natl. Acad. Sci. U.S.A. 110, 1175-1180.

65

Manokaran N. and LaFrankie J.V. (1990) Stand structure of Pasoh Forest Reserve, a lowland rainforest in peninsular Malaysia J. Tropical Forest Sci. 3, 14-24.

Mariotti A. and Peterschmitt E. (1994) Forest savanna ecotone dynamics in India as revealed by carbon isotope ratios of soil organic matter. Oecologia 4, 475-480.

McClaren M.P. and McPherson G.R. (1995) Can soil organic carbon isotopes be used to describe grass-tree dynamics at a savanna-grassland ecotone and within the savanna? J. of Vegetation Sci. 6, 857-862.

Medina E. and Minchin P. (1980) Stratification of δ13C values of leaves in Amazonian rainforests. Oecologia 45, 377-378.

Meir P., Kruijt B., Broadmeadow M., Barbosa E., Kull O., Carswell F., Nobre A. and Jarvis P.G. (2002) Acclimation of photosynthetic capacity to irradiance in tree canopies in relation to leaf nitrogen concentration and leaf mass per unit area. Plant, Cell and Environment 25, 343-357.

Morley R.J. (2000) Origin and Evolution of Tropical Rainforests. John Wiley & Sons, Chichester UK.

Mueller K.E., Polissar P.J., Oleksyn J. and Freeman K.H. (2012) Differentiating temperate tree species and their organs using lipid biomarkers in leaves, roots, and soils. Org. Geochem. 52, 130-141.

Nadelhoffer K.J. and Fry B. (1988) Controls on natural nitrogen-15 and carbon-13 abundances in forest soil organic matter. Soil Sci. Soc. of Am. Journal 52, 1633- 1640.

Niinemets Ü., Kull O. and Tenhunen J.D. (1999) Variability in leaf morphology and chemical composition as a function of canopy light environment in coexisting deciduous trees. Int. J. Plant Sci. 160, 837-848.

Niinemets Ü. (2007) Photosynthesis and resource distribution through plant canopies. Plant, Cell and Environment 30,1052-1071.

Odum H.T. (1970) Summary: an emerging view of the ecological system at El Verde. In: A Tropical Rainforest (eds. H.T. Odum and R. Pigeon) Division of Technical Information, U.S. Atomic Energy Commission, Washington D.C. pp. I191-I289.

Ometto, J.P.H.B., Ehleringer J.R., Domingues T.F., Berry J.A., Ishida F.Y., Mazzi E., Higuchi N., Flanagan L.B., Nardoto G.B. and Martinelli L.A. (2006) The stable carbon and nitrogen isotopic composition of vegetation in tropical forests of the Amazon Basin, Brazil. Biogeochem. 79, 251-274.

66 Osada N., Takeda H., Furukawa A. and Awang M. (2001) Leaf dynamics and maintenance of tree crowns in a Malaysian rainforest stand. J. Ecol. 89, 774-782.

Ozanne C.M.P., Anhuf D., Boulter S.L., Keller M. and Kitching R.L. (2003) Biodiversity meets the atmosphere: global view of forest canopies. Science 301, 183–187.

Parker G.G. (1995) Structure and Microclimate of Forest canopies. In: Forest Canopies (eds. M.D. Lowman and N.M. Nadkarni). Academic Press Inc., London UK. pp. 73–106.

Parker G.G., O’Neill J.P. and Higman D. (1989). Vertical profile and canopy organization in a mixed deciduous forest. Vegatatio 85, 1-11.

Pataki D.E., Ehleringer J.R., Flanagan L.B., Yakir D., Bowling D.R., Still C.J., Buchmann N., Kaplan J.O. and Berry J.A. (2003) The application and interpretation of Keeling plots in terrestrial carbon cycle research. Global Biogeochem. Cycles 17, 1022-1029.

Phillips D.L. and Koch P.L. (2002) Incorporating concentration dependence in stable isotope mixing models. Oecologia 130, 114-125.

Polgar C.A. and Primack R.B. (2011) Leaf-out phenology of temperate woody plants: from trees to ecosystems. New Phytologist 191, 926-941.

Poorter H., Pepin S., Rijkers T., de Jong Y., Evans J.R. and Körner C. (2006) Construction costs, chemical composition and payback time of high- and low- irradiance leaves. J. Exp. Botany 57, 355-371.

R Development Core Team, 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-9000051-07-0, URL http://www.R-project.org.

Rasse D.P., Rumpel C., and Dignac M.F. (2005) Is soil carbon mostly root carbon? Mechanisms for a specific stabilisation. Plant and Soil 269, 341-356.

Reich P.B., Uhl C., Walters M.B. and Ellsworth D.S. (1991) Leaf lifespan as a determinant of leaf structure and function among 23 Amazonian tree species. Oecologia 86, 16-24.

Richards P.W. (1996) The Tropical Rainforest: An Ecological Study. (2nd ed.) Cambridge University Press, Cambridge UK.

Santiago L.S. (2003) Leaf traits of canopy trees on a precipitation gradient in Panamá: Integrating plant physiological ecology and ecosystem science. Ph.D. thesis, University of Florida.

67 Santiago L.S. and Wright S.J. (2007) Leaf functional traits of tropical forest plants in relation to growth form. Functional Ecol. 21,19-27.

Secord R., Wing S.L. and Chew A. (2008) Stable isotopes in early Eocene mammals as indicators of forest canopy structure and resource partitioning. Paleobiol. 34,282- 300.

Spicer R.A. (1980) The importance of depositional sorting to the biostratigraphy of fossil plants. In: Biostratigraphy of Fossil Plants in successional and paleoecological analyses (eds. D.L. Dilcher and T.N. Taylor). Hutchinson & Ross Inc., Stroudsburg, PA. pp. 171–183.

Sternberg L.D.L., Mulkey S.S. and Wright S.J. (1989) Ecological interpretation of leaf carbon isotope ratios: influence of respired carbon dioxide. Ecology 70, 1317- 1324.

Stewart G.R., Turnbull M.H., Schmidt S. and Erskine P.D. (1995) 13C natural abundance in plant communities along a rainfall gradient: a biological integrator of water availability. Aust. J. Plant Physiol. 22, 51-55.

Sussman R.W. (2005) Primate origins and the evolution of angiosperms. Am. J. Primatol. 23, 209-223.

Tiffney B.H. (1984) Seed size dispersal syndromes, and the rise of the angiosperms: Evidence and hypothesis. Ann. MO Botanical Garden 71, 551-576.

Upchurch G.R. and Wolfe J.A. (1987) Mid-Cretaceous to early Eocene vegetation and climate: evidence from fossil leaves and woods. In: The Origins of Angiosperms and their Biological Consequences (eds. E.M. Friis, W.G. Chaloner and P.R. Crane). Cambridge University Press, Cambridge UK. pp. 75-105. van der Merwe N.J. and Medina E. (1989) Photosynthesis and 13C/12C ratios in Amazonian rainforests. Geochim. Cosmochim. Acta 53, 1091–1094. van Dijk A.I.J.M. and Keenan R.J. (2007) Planted forests and water in perspective. Forest Ecol. and Management 251, 1–9.

Vogel J.C. (1978) Recycling of carbon in a forest environment. Oecologia Plantarum 13, 89–94.

Wilf P. (2000) Late Paleocene-early Eocene climate changes in southwester Wyoming: paleobotanical analysis. Geolog. Soc. of Am. Bull. 112, 292-307.

Williams C.J., Johnson A.H., LePage B.A., Yann D.R. and Sweda T. (2003) Reconstruction of Tertiary Metasequoia forests. II. Structure, biomass, and

68 productivity of Eocene floodplain forests in the Canadian Arctic. Paleobiol. 29, 271-292.

Wills C. and Condit R. (1990) Similar non-random processes maintain diversity in two tropical rainforests. Proc. R. Soc. London, Ser. B 266, 1445-1452.

Wilson E.O. (1994) Naturalist. Island Press, Washington, DC.

Wing S.L., Bown T.M. and Obradovich J.D. (1991) Early Eocene biotic and climatic change in interior western North America. Geology 19, 1189-1192.

Wing S.L., Bao H.M. and Koch P.L. (2000) An early Eocene cool period? Evidence for continental cooling during the warmest part of the Cenozoic. In: Warm Climates in Earth History (eds. B.T. Huber, K. MacLeod and S.L. Wing). Cambridge University Press, Cambridge UK. pp. 197-237.

Wing S.L., Herrera F., Jaramillo C.A., Gómez-Navarro C., Wilf P. and Labandeira C.C. (2009) Late Paleocene fossils from the Cerrejón Formation, Colombia, are the earliest record of Neotropical rainforest. Proc. Natl. Acad. Sci. U.S.A.106, 18627- 18632.

Wright I.J. and Cannon K. (2001) Relationships between leaf lifespan and structural defenses in a low nutrient, sclerophyll flora. Functional Ecol. 15, 351-359.

Wright S.J., Jaramillo M.A., Pavon J., Condit R., Hubbell P.B. and Foster R.B. (2005) Reproductive size thresholds in tropical trees: variation among individuals, species and forests. J. Tropical Ecol. 21, 307–315.

Wynn J.G (200) Carbon isotope fractionation during decomposition of organic matter in soils and paleosols: Implications for paleoecological interpretations of paleosols. Palaeogeography, Palaeoclimatol., Palaeocol. 251, 437–448.

Zimmerman J.K. and Ehleringer J.R. (1990) Carbon isotope ratios are correlated with irradiance levels in the Panamanian orchid Catasetum viridiflavum. Oecologia 83, 247-249.

69 Chapter 3. Biogeochemical confirmation of the first closed-canopy Neotropical forest in the fossil record: A novel application of isotopic methods

3.1. Abstract

We evaluated and interpreted canopy coverage based on the bulk δ13C compositions of

199 fossil leaves from the Guaduas Formation (Cretaceous) and from two separate sites within the Cerrejón Formation (Paleocene), both in modern-day Colombia. The Guaduas fossils are among the only angiosperm leaves found in the Maastrichtian Neotropics and the Cerrejón paleoflora is thought to represent the earliest closed-canopy rainforest in the

13 Neotropics. The range of δ Cleaf values for leaves from the Guaduas Formation and one site in the Cerrejón Formation are consistent with an open-canopy structure. The other

13 Cerrejón site expressed the larger δ Cleaf range consistent with a closed-canopy forest.

The lithology and taxa in the open-canopy Paleocene site indicate a lacustrine depositional environment and it’s assemblage was likely biased towards leaves from the

13 lake edge. Under such conditions the canopy leaves would not express the wide δ Cleaf associated with a fully closed canopy. Although open-canopy deposits are likely to be more common in the fossil record but our isotopic analyses suggest a closed-canopy condition for the other Paleocene site, confirming the paleobotanical evidence that the

Cerrejón paleoflora is the first example of a Neotropical forest in the fossil record.

70 3.2. Introduction

The rise of closed-canopy forests is often linked to the widespread distribution of angiosperms in the Paleocene, but the location and age of first forest canopies is disputed

(Germeraad et al., 1968; Upchurch & Wolfe, 1987; Spicer et al., 1993; Morley, 2000;

Johnson & Ellis, 2002; Burnham & Johnson, 2004; Davis et al., 2005; Jaramillo et al.,

2010). Closed-canopy forest ecosystems support shade-tolerant taxa that are adapted to take advantage of gap formation (Lang & Knight, 1983). Although shade tolerance is not unique to the angiosperms, 80-90% of species in extant closed-canopy tropical forests are angiosperms (Burnham & Johnson, 2004; Valladares & Niinemets, 2008; Jaramillo et al.,

2010).

The Cerrejón paleoflora of the Late Paleocene is the earliest fossil example of a rainforest in the Neotropics (Wing et al., 2009). Most leaves from this site represent the angiosperm families associated with modern rainforests, including many palms and legumes. Leaf architecture indicates very high (>2500 mm/yr) rainfall and there is evidence of highly diverse herbivorous insects (Wing et al., 2009). Even though the

Cerrejón flora has characteristics consistent with modern tropical rainforests, it is difficult to reconstruct canopy closure from fossil data alone. An older paleoflora from the

Maastrichtian Guaduas Formation contains dicotolydenous leaves with physiognomic features typical of tropical environments but lacks key floristic elements of a closed- canopy forest. The Guaduas site has a gymnosperm pollen assemblage that does not have any modern analog (Jaramillo et al., 2010; Jaramillo, personal communication).

71 Information about canopy closure will help document the links between angiosperm evolution and the development of three-dimensional ecologic structure in tropical rainforests. Yet, it is difficult to gauge forest density in paleofloras by comparison to modern forests. Early forests are not directly analogous to modern forests since fossil assemblages may include extinct and extant taxa that are highly cosmopolitan.

The origin and evolution of closed-canopy forests is of interest to climatologists and paleoecologists alike. Closed-canopy forests have significant effects on regional and global climate by regulating local conditions to maintain the warm, moist parameters to which they are best adapted (Betts et al., 2007). Closed-canopy tropical forests thrive in year-round warmth (> 20°C) and high precipitation (> 2000 mm/year) regimes (Turner,

2001). The extent of evapotranspiration in the tropical rainforests increases surface moisture availability, which in turn, regulates global temperature, hydrologic cycling and atmospheric circulation (Bastable et al., 1993; Betts, 1999; Boyce et al., 2010). Tropical rainforests are only ~12% of the Earth's surface but account for 45% of carbon in terrestrial biomass (IPCC, 2000; Grace et al., 2001; Malhi et al., 2002).

Leaves exhibit morphological and physiological features that reflect their light environment. Shade leaves are typically larger and thicker, with greater amounts of chlorophyll and nitrogen but fewer stomata (Valladares & Niinemets, 2008). These leaves have slower rates of photosynthesis but are longer-lived than leaves in the well-lit canopy (Bongers & Popma, 1990; Reich et al, 1991; Santiago & Wright, 2007; also see

Valladares & Niinemets, 2008 and references therein). Some of these features are

72 measurable in fossil leaves yet positive identification of stand architecture from fossil leaves is difficult. Morphological differences can be very subtle and require exceptional preservation (Burnham et al., 1992; Ellis et al., 2009). Paleo-canopy reconstructions typically rely on physiognomic features of fossil leaves as an indicator of climate in addition to the presence of taxa associated with extant closed-canopy forests (Johnson &

Ellis, 2002; Davis et al., 2005; Wing et al., 2009). A geochemical proxy that represents a biogeochemical metric of light environment would complement physiognomic leaf analyses. A biomarker approach could also open opportunities to study canopy structure at sites lacking well-preserved fossils.

Leaves sampled vertically through closed-canopy forests exhibit a large range (up to

13 ~10‰) in δ Cleaf from canopy top to understory (Medina & Minchin, 1980; Ehleringer et al., 1986; Lockheart et al, 1997; Meir et al., 2002; Graham et al., in review). By

13 comparison, the δ Cleaf range in temperate seasonal forests is more modest, about 6‰

(Buchmann et al., 1997; Graham et al., in review).

13 13 The C-abundance in bulk leaf tissue (expressed as δ Cleaf) declines with decreasing canopy height. This “canopy effect” is a long-recognized biochemical indicator of canopy conditions in extant leaves and reflects co-variation of light, humidity, and the isotopic composition of atmospheric CO2 (Vogel, 1978). Higher carbon assimilation rates in the well-lit canopy top cause leaves to draw down internal CO2 rapidly, thus lowering

13C discrimination when compared to leaves in the shaded understory (Ehleringer et al.,

1986; Zimmerman & Ehleringer, 1990; Hanba et al., 1997; Lockheart et al., 1997;

73 Heaton, 1999; Niinemets et al., 1999). The two- to four-fold increase in photosynthetic rate in upper compared to lower canopy leaves contributes to the isotopic canopy effect

(Baldocchi, 1993; Ellsworth & Reich, 1993).

If leaves capture the isotopic distinction between canopy top and shaded understory, then carbon isotope analysis of fossil leaves could provide a proxy for canopy cover. Leaves with higher photosynthetic rates, including those in the well-lit upper canopy, are relatively short-lived (Reich et al., 1991). This leads to a very high relative flux of leaves from the upper canopy strata going into litter assemblages and, by extension, into the plant fossil record. The preservation of understory leaves is more rare. In Graham et al.,

13 (in review) we employed a Monte Carlos style resampling model that combined δ Cleaf and leaf biomass data to understand the probability of finding diagnostic understory leaves in a random sample. We found that a minimum collection of 50 leaves from a paleofloral assemblage are required to capture a carbon isotope range that would indicate canopy closure.

In this study, we analyzed the carbon isotope composition of fossil leaves from the Late

Paleocene Cerrejón and Latest Cretaceous Guaduas paleofloras. The goal was to

13 determine the range of δ Cleaf values as a proxy for the extent of canopy closure in tropical vegetation before and after the K-Pg extinction which has been hypothesized to affect forest structure and diversity (Johnson, 1992; Wing & Boucher, 1998; Wilf &

Johnson, 2004; Wilf et al., 2006; de la Parra et al., 2007). Our sampling plan and

74 comparison data were based on the modern forest data evaluated with a resampling model (Graham et al., in review).

The flora of the Cerrejón include many of the dominant families in closed-canopy forests of the modern Neotropics. Macrofossils used in this study include representatives of the

Fabaceae, Moraceae, Annonaceae, Euphorbaceae, Lauraceae, Sapotaceae, Myristicaceae, and Arecaceae, amongst others (Herrera 2008a; Herrera 2008b; Wing et al., 2009). These families represent nearly 75% of the diversity and almost 60% of the stems in a modern analog forest (see Burnham & Johnson, 2004 and references therein; Jaramillo &

Cardenas, 2013). The Cerrejón flora also include many fossil leaves allied with lianas in the family Menispermaceae – a family and growth habit ubiquitous in closed-canopy

Neotropical forests (Doria, et al., 2008). Leaf physiognomic features– size, entire margins, vein density –suggest a closed, multi-stratal rainforest (Wing et al., 2004;

Herrera, 2008a; Jaramillo et al., in prep).

Most angiosperm leaves from the Guaduas flora are not assigned to families and are described as indeterminate dicotyledenous. The few angiosperm leaves and pollen that do represent familiar modern taxa have been assigned to the families Rhamnaceae,

Piperaceae, Lauraceae, Urticaceae, and Arecaceae and the order Canellales (Correa et al.,

2010, F. Herrera, personal communication). While these families are typical of the modern Neotropics, the dominant modern taxa such as Fabacaeae, Annonaceae, and

Euphorbaceae are missing (Burnham & Johnson, 2004; Wing et al., 2009). Given the pollen evidence for abundant gymnosperms and the non-analog distribution of

75 angiosperm families, it is difficult to determine how closely this flora resembled a modern closed-canopy tropical forest. Communities that may have been common in the

Cretaceous would have no modern counterparts, due largely to the major floral turnover associated with the K-T extinction (Wilf et al., 2006; de la Parra et al., 2007).

3.3. Methods

3.3.1. Collection sites

Maastrichtian Assemblage

The Guaduas flora - from the middle Guaduas Formation of Boyaca department in central

Colombia (see Fig. 1) - remains one of the only sources of Maastrichtian leaf fossils in

South America and the northernmost fossil plant locality in the Cretaceous record of

South America. Fossil leaves were collected at 5˚55'45"N, 72˚47’43"W. The paleolatitude of this site is closer to equator at 2˚N (Sarmiento, 1992; Correa et al., 2010).

Reconstruction of mean annual precipitation (MAP) and mean annual temperature

(MAT) from leaf size and leaf margin analysis suggest an MAP of ~2400 mm/year and an MAT of 22.1 +/- 3.4˚C (Gutierrez & Jaramillo, 2007). Palynofloral assemblages from

Guaduas can be assigned to palynological zones that indicate an age of ca. 68-66 Ma

(Muller et al., 1987; Jaramillo & Rueda, 2004; Correa et al., 2010). Specimens used in this study are housed in the paleontological collection of the Center for Tropical

76 Paleoecology and Archeology at the Smithsonian Tropical Research Institute in Panamá

City, Panamá . This paleoflora is a recent discovery and complete description of leaves is still in preparation. However, some macrofossils have been described in Gutierrez &

Jaramillo (2007) and Correa et al., (2010).

The Guaduas formation is well described and is actively mined for coal. The lower units of the Guaduas formation where macrofossils are found are fine-grained sand beds intercalated with coal seams. These strata are overlain conformably by laminated and massive mudstones with sandstone interbeds. The Guaduas formation contains the K-T boundary and is beneath the Paleocene-lower Eocene Cacho and Bogota Formations. The

Guaduas Formation, now 2700 m above sea level, was deposited in a coastal plain

(Guerrero, 2002).

Paleocene Assemblages

The Cerrejón flora is found in the Cerrejón Formation, which crops out throughout the

Guajira Peninsula of northern Colombia and, most notably, is exposed in an open pit coal mine at the base of the peninsula (see Fig. 1). Samples for this study were taken from within the mine from two vertically separated beds located at 11˚5'60"N, 72˚30’0"W and are described in Wing et al. (2009). Specimens are housed by the Smithsonian Institution either at the National Museum of Natural History in Washington DC or at the Center for

Tropical Paleocology and Archeology at the Smithsonian Tropical Research Institute in

Panama City, Panama .

77

The paleolatitude of this location is 5˚30'N (Jaramillo et al., 2007). Leaf morphology and size suggest MAT was 24 - 31˚C. and MAP of 2,260-4,640 mm/year (Wilf et al., 1998;

Herrera et al., 2008a; Wing et al., 2009). These data support other work indicating a warmer, wetter Paleocene climate when compared to the late Cretaceous (Leidelmeyer,

1966; Parrish, 1998; see Burnham & Johnson, 2004 and references therein).

Palynological assemblages indicate ages of middle-to-late Paleocene, ~58-60 Ma, but the

Formation is nearly one kilometer thick, implying very high rates of sedimentation

(Bayona et al., 2004; Jaramillo et al., 2007). This is also evident in the wide variety of lithologies (sandstones, mudstones, and coals) in the Cerrejón Formation. Sediments in the basin suggest a mosaic of fluvial and lacustrine settings typical of an estuarine coastal plain which, based on deposition rates, was likely subject to frequent flooding, avulsions, and channel migration (Jaramillo et al., 2007). Leaf fossils for this study were taken from sites 0315 and 0318 (see Fig. 2). Both sites were constrained to thin, yet productive, beds with lateral areas of only 4-6 m2.

Site 0315 is dominated by Fabaceae (42% of all macrofossils). Except for one specimen, all Fabaceae leaves at site 0315 are the same morphotype. Fabaceae, Lauraceae, and

Sapotaceae, the most common families, comprise 68% of the morphotypes identified at this site. Fossils at site 0315 are found in heterolithic sediments suggestive of low-to- medium energy channels. Plant remains are found in fine-grained, massive to laminated mudstones and siltstones. These lens-shaped mudstones and siltstones within inclined

78 strata are cut through siltstones are by massive to cross-bedded sandstones indicating a channel cutting through a coastal plain (Jaramillo et al., 2007; Wing et al., 2009).

Site 0318 is, stratigraphically, 400 m above site 0315, or approximately one million years younger based on sedimentation rates (Bayona et al., 2004; Jaramillo et al., 2007). This site is co-dominated by Malvaceae and Fabaceae, which together comprise 40% of the morphotypes. The Fabaceous morphotypes at 0318 are different from those at 0315. The most common families among morphotypes at this site are Fabaceae, Malvaceae,

Arecaceae, Menispermaceae, and Violaceae. These families represent 74% of relative diversity. Site 0318 is one of ten fossil sites in the Cerrejón Formation that are found in flat-laminated, thinly-bedded, tabular siltstones, interpreted as small lakes (Wing et al.,

2009).

3.3.2. Sampling and analytical methods

Fossil Sampling

Small portions of cuticle and mesophyll were removed from fossil leaves from the

Guaduas and Cerrejón paleofloras. Samples were identified by inspection and compared with morphotype descriptions from Herrera et al. (2008a), Wing et al. (2009), and Correa et al. (2010). Prior to this study, rocks had been split along bedding planes and dusted with a soft bristle brush to reveal leaf structure and morphology. Rock material adjacent to the fossils was scraped or ground away in order to reveal leaf edges. The surfaces of

79 the fossils were kept as clean as possible in order to prevent damage or obfuscation of features. Prior to sampling for carbon isotope analyses, fossils were lightly brushed and, when possible, gently rinsed with deionized water to remove any surface contaminants.

Fossil leaves were selected based on preservation quality and whether or not they had been assigned to an extant family.

For this study, 68 leaf samples were taken from Cerrejón site 0315. Specimens included

10 morphotypes from nine families, as well as a selection of indeterminate dicot leaves

(leaves not assigned to a morphotype). 78 leaf samples were taken from Cerrejón site

0318. Specimens included 19 morphotypes, from 10 families and a selection of indeterminate dicot leaves. Of the 53 leaf samples taken from the Guaduas flora only two were identified to the level of family, 41 were indeterminate dicots, and the remaining 11 samples were assigned to one of five morphotypes.

Fossil Cleaning and Pre-Treatment

Compression fossils from both localities have preserved carbonized leaf material (i.e. phytoleim) the remains, presumably, of cuticle and mesophyll. Organic material was picked, scraped, or ground from the rock matrix using either forceps, scalpels or a rotary grinding tool equipped with a diamond-pointed bit, respectively. Every effort was made to minimize extraneous material in cuticle samples. Cuticle that could be removed by forceps or scalpel had sediment brushed from it’s surface. Fossils that required vigorous scraping or grinding to remove cuticle or mesophyll were sampled under a dissecting

80 microscope to ensure that a minimum of sediment was introduced into the sample. All samples were crushed and homogenized with a Teflon pestle. Samples were then treated with 6N HCl for two hours, and then rinsed three times with deionized water. This treatment was repeated until the final rinse water maintained a neutral pH. Samples were lyophilized before isotopic analysis.

Stable Isotope Analysis

Carbon isotope analyses were performed at Pennsylvania State University using a

Costech elemental analyzer connected to a Thermo Finnigan Delta XP IRMS. Analyses were performed under continuous-flow (He; 120 ml/min), oxidized at 1020°C over chromium (III) oxide and cobaltous silver (II, III) oxide, then reduced over elemental copper at 650°C before passing through a water trap and into IRMS. δ13C values were corrected for sample size and reported relative to Vienna Pee Dee Belemnite (VPDB)

(Coplen et al., 2006). Error was determined by repeated analysis of internal standards

(calibrated using IAEA CH-7, IAEA CH-6, and USGS-24; polyethylene, sucrose and graphite, respectively). Instrument precision is ±0.09‰ (n=76). Accuracy, the average difference between the measured and true δ13C value, is 0.02‰ (n = 96). Duplicate analyses were performed for 50% of all samples and samples that defined the end points of the isotopic range were also analyzed in triplicate.

3.4. Results

81 13 The average δ Cleaf value for both of the Cerrejón sample sets was -25.9‰. The younger flora from site 0318 expressed a range of 3.3‰, from -26.9 to -23.6‰ while the older

13 flora at site 0315 had a range of 6.3‰, from -22.7 to -29.0‰. The average δ Cleaf value for the Guaduas collection was -26.0‰. Leaves in the Guaduas flora expressed a range of

2.7‰, from -27.5 to -24.7‰ (see Fig. 3).The 13C enrichment of these leaves (when compared to modern C3 plants) is consistent with the higher δ13C values estimated for carbon dioxide in the atmosphere at that time (Tipple et al., 2010). The isotopic composition of atmospheric CO2 during those time intervals was +3.3 to 3.8‰ greater than the value for atmospheric CO2 in 2010 (-8.2‰; UCSD Scripps Global CO2 Program,

2012).

∆leaf is the fractionation of carbon during photosynthesis. The isotopic signal reflects the concentration of CO2 within leaf stomata under conditions of constant atmospheric CO2.

Defined by Farquhar et al. (1989)

13 13 13 ∆leaf = (δ Catm - δ Cleaf)/(1 + δ Cleaf / 1000)

Internal leaf CO2 concentration is regulated by stomatal conductance and the flux of carbon fixed into photosynthate. Water stress leads to a closing of stomata which leads to a decrease in ∆leaf. Hence, well-lit and water-stressed leaves have smaller ∆leaf values and leaves growing under high humidity and low irradiance have the most positive ∆leaf values.

82 13 Using reconstructed δ C values for atmospheric CO2 (-4.4‰ for Paleocene and -4.9‰ for Cretaceous) from Tipple et al., (2010), we calculated the carbon isotope discrimination (∆leaf) of leaves. These values are compared with the ∆leaf values of leaves collected at the Panamá site. Fractionation factors for leaves from all three paleoflora are within the ∆leaf range observed for modern tropical rainforests (Diefendorf et al., 2010).

3.5. Discussion

None of the three floras had normally distributed δ13C values (Fig. 4). The distributions are skewed toward more enriched values, suggesting a greater occurrence of 13C-enriched mid- and upper-canopy leaves and a relative rarity of 13C-depleted understory leaves.

Greater than 55% of leaf biomass is produced in the mid- to upper canopy in any given forest type (Graham et al., in review). Further, transport processes favor the movement of canopy leaves into depositional environments, biasing the fossil record towards leaves of the upper canopy (Ferguson, 1985; Spicer, 1989).

We compared measured δ13C ranges for each leaf assemblage to the results of modern forests from Panama and Maryland (Graham et al., in review). These modern forest ranges were determined using a model that randomly sampled leaf carbon isotope values from a canopy data set. Selection probability was set to account for differences in biomass flux from different canopy elevations. The most probable range for a fossil leaf collection of a particular size was calculated from the virtual collection. The smallest sample from the Guaduas Formation (n = 53) would have a minimum range of 4.9‰ if it

83 were derived from a closed-canopy forest. Leaves from the Cerrejón flora at site 0315 (n

= 68) would have a minimum range of ~5.2‰, and Cerrejón site 0318 (n = 78) had a minimum range of ~5.4‰ under closed-canopy conditions. Of the three paleofloras analyzed, only Cerrejón site 0315 had an isotopic range (6.3‰) consistent with a closed- canopy forest (see Fig. 5).

3.5.1. Guaduas (Maastrichtian)

The mix of gymnosperm pollen and angiosperm macrofossils in the Guaduas Formation does not have a modern analog. Further, many of the fossils cannot be assigned to any extant plant families. This makes modern-to-ancient floristic comparisons difficult and we cannot constrain canopy closure using modern analogs. The leaf carbon isotope range

(2.7‰) is very narrow and this strongly suggests an open canopy structure (Graham et al., in review). For example, an isotopic canopy effect this small is found in open woodlands such as a South African savanna (Codron et al., 2005).

The projected MAT and MAP reconstructions and ∆leaf values for the Guaduas flora are consistent with the modern tropical rainforest, but the isotopic range is not (Fig. 6). The narrow isotope range could indicate a deciduous tropical forest. Deciduous tropical forests have highly seasonal rainfall and exhibit leaf senescence during dry periods, resulting in open canopies during leaf flush. Lacking key floristic components, the composition of the Guaduas flora approaches, but is not analogous to, either modern seasonally deciduous tropical forests or tropical rainforests. Although seasonally wet or

84 disturbed tropical forests often have abundant pteridophytes and gymnosperms today, they are dominated by angiosperms (Burnham & Johnson, 2004). While pollen and macrofossils from different plant communities can be idiosyncratically preserved together there is no evidence of reworking at this site (Farley, 1989).

Leaf margin analysis indicates a climate with high MAP while the δ13C range for those same leaves suggest an open canopy structure but with so few leaves taxonomically identifiable and the non-analog mix of floristic components it is difficult to make comparisos to modern ecosystems in order to understand structure.

3.5.2. Cerrejón Site 0315

The range of δ13C values for leaves at site 0315 is 6.3‰, and this falls within the modeled range for leaves from a closed-canopy forest. This finding is consistent with the suggestion that this was the first closed-canopy forest of the Neotropics (Wing et al.,

2009). Combined with paleoclimatic parameters, leaf physiognomic features, and the dominance of taxa associated with modern closed-canopy forests, this assemblage has many characteristics considered to be typical of the tropical rainforest.

Closed-canopy conditions in modern forests are generally associated with high diversity.

The Cerrejón paleoflora is less diverse than is typical of a modern tropical rainforest, even thought the isotopic range is suggestive of closed-canopy and the floristic elements are common in modern closed-canopy forests (Wing et al., 2009). This low diversity in

85 the paleoflora could actually be due to closed canopy conditions. In temperate deciduous forests, leaf litter represents only a limited number of species since it is dominated by leaves from canopy trees (Burnham et al., 1992; (Ferguson, 1985). Fossil leaf assemblages are likely subject to the same taphonomic biases as collections of leaf litter.

Alternately, lower diversity, closed-canopy conditions can be found in modern tropical wetlands. Paleocene sediments in this area typically include an abundance of pollen from

Nypa, a mangrove palm typical of tropical shorelines (Jaramillo & Dilcher, 2000). Leaves and an inflorescence from the Nypa were found at site 0318 in the Cerrejón Formation, but no fossils associated with Nypa are found at site 0315. While the depositional setting certainly reflects a wet, coastal plain site 0315 may not be a tropical wetland ecosystem.

The low diversity of the Cerrejón paleoflora combined with closed-canopy may indicate that the development of a closed-canopy structure preceded the high speciation familiar in modern tropical rainforests. In other words, plant diversification or immigration may have followed after canopy closure, by fostering the innovation of juvenile shade tolerance and other strategies that allowed multi-stratal forests to develop (Turner, 2001;

Myers & Kitajima, 2007). Low diversity may also reflect a long recovery from the K-T extinction, regardless of the closed-canopy coverage (la Parra et al., 2007).

3.5.3. Cerrejón Site 0318

Site 0318 has many markings of a closed-canopy tropical forest. Site 0318 has High

MAT and MAP, large entire-margin leaves, and many families associated with extant

86 closed-canopy forests. In contrast, the carbon isotope range of leaf material is far below the expected range for a random sampling of closed-canopy forest leaves. Model results show that for a sample of this size the leaf carbon isotopic range would be at least 5.2‰ if the leaves were from a closed-canopy forest (Graham et al., in review). The range at site 0318 was only 3.3‰.

Leaves from site 0318 were found in an interbedded lens-shaped body of siltstone and sandstone (Wing et al., 2009). This indicates a low-to-medium energy channel that might have that captured canopy leaves from upstream as well as the immediate vicinity of the deposit (Spicer, 1981; Spicer & Wolfe, 1987). Leaves from site 0318 were found in a tabular bed of flat, thinly-bedded, gray siltstone. This type of sedimentation suggests a small, low-energy, lacustrine environment. Autochthonous deposits such as this capture primarily leaves from the lake margin vegetation and are biased toward the smaller sun leaves that flourish at the lake edge and against larger shade leaves (Spicer, 1981; Spicer

& Wolfe, 1987; Burnham et al., 1992; Greenwood, 1992; Greenwood., 2007).

Taxa found at site 0318 are consistent with a lacustrine environment, and include abundant menisperm lianas and palms (Wing et al., 2009; Herrera et al., 2011). Lianas tend to proliferate in gaps, disturbed areas, or at lake edges, constituting up to 1/3 of leaf biomass in such settings (Rice et al., 2004). Thus, sedimentary and floristic evidence point toward an autochthonous collection of leaves from a primarily open lake edge. This setting is consistent with isotopic evidence for open canopy structure. The model from which the carbon isotope range expectations were developed considered leaves in the full

87 range of light environments that would be possible in the interior of a forest, essentially a random sampling of leaves from the litter bed. An allochthonous fluvial assemblage integrating a wider area and a greater variety of leaves, both sun and understory, will more resemble this type of collection (Spicer & Wolfe, 1987). The fossil sites from

Guaduas and Cerrejón were not strictly in situ litter collections, although, given the siltstone lithology and evidence of channel formation, leaves were probably not transported for great distances. Modern leaves along exposed forest edges can have a leaf

δ13C composition 2‰ more enriched than leaves from the same species and height above ground just 50 m into a forest from the edge (Kapos et al., 1993). The resampling model did not account for the abundance of edge leaves in paleofloral assemblages (Graham et al., in review).

Fossil assemblages commonly represent vegetation adjacent a stream or body of water with generally open-canopy conditions. Sedimentological evidence from both the

Cerrejón sites and the Guaduas site suggest that originating vegetation were not exclusively from the interior of a forest. Assessments of canopy closure in fossil assemblages would be, by default, open-canopy, making the capture of a closed-canopy isotopic signal in leaves from Cerrejón site 0315 a remarkable finding.

3.6. Conclusions

13 δ Cleaf in fossil leaves are consistent with paleobotanical evidence of canopy closure.

Isotopic analysis, combined with floristic analysis, can clarify the vegetation habit likely

88 in ambiguous, poorly preserved or cosmopolitan fossil taxa. Our results suggest that

Cretaceous flora in the Guaduas formation did not grow in closed-canopy, multi-stratal conditions, despite the climatic similarity and floristic affinity with modern tropical rainforests. Our methods confirmed that for site 0315 the Cerrejón flora is a closed- canopy, tropical forest and therefore represents the first of it’s kind in the Neotropics. Our methods also confirmed that Cerrejón site 0318 was likely a more open canopy, as would be expected for vegetation along a lake edge. Isotope data, with floristic and lithologic insights, were able to distinguish an allochthonous deposit from a forests edge assemblage. When combined with floristic and physiognomic analysis of leaf material carbon isotope analysis of preserved cuticle and mesophyll can prove to be a powerful indicator of canopy structure in ancient forests.

89 3.7. Figures

Figure 3.1. Modern leaf collection site in Panamá site and three Guaduas and Cerrejón fossil collection sites in two locations in Colombia. Digital elevation model by Amante & Eakins (2009).

90

Figure 3.2. Stratigraphic column of Cerrejón Formation fossil beds from Wing et al., (2009). Fossils were selected from assemblages 0315 and 0318, vertically 100 m apart.

91 13 Figure 3.3. Measured δ Cleaf range for all modern and fossil leaves. The box spans the interquartile range with the median denoted within the box and the whiskers are defined by the upper and lower quartiles.

92

13 Figure 3.4. Histograms of measured δ Cleaf range for fossil leaves show the frequency 13 13 with which δ Cleaf values are expressed in fossil leaves and the rarity of C-depleted understory leaves. The right skew in the data is likely due to the bias in fossil assemblages towards preservation of enriched upper canopy leaves.

93

13 Figure 3.5. Absolute δ Cleaf range for each fossil leaf site compared with the predicted isotopic range for closed canopy litter based on sample size from Graham et al. (in review; see Chapter 2)

94

Figure 3.6. Calculated ∆leaf values for modern leaves from three binned strata of the forest and ∆leaf values for fossil leaves from three study localities. Lower ∆leaf values area associated with upper canopy leaves and all fossil sites express ∆leaf values more typical of upper canopy leaves.

95 3.8. References

Baldocchi D.D. (1994). A comparative study of mass and energy exchange over a closed (wheat) and an open (corn) canopy: I. the partitioning of available energy into latent and sensible heat exchange. Agricultural and Forest Meteorol. 67, 191-220.

Bastable H.G., Shuttleworth J., Dallarosa R.L.G., Fisch G. and Nobre C.A. (1993) Observations of climate, albedo, and surface radiation over cleared and undisturbed Amazonian forest. Int. J.Climatol. 13, 783-796.

Bayona G., Jaramillo C.A. Rueda M. Pardo A. and Hernandez G. (2004) Important paleotectonic and chronostratigraphic considerations of the Late Paleocene in the northernmost Andes as constrained by Paleocene rocks in the Cerrejón Coal Mine, Guajira, Colombia. III Convención Técnica Asociación Colombiana de Geólogos y Geófisicos del Petroleo, Bogotá, Colombia.

Betts R.A., Falloon P.D., Goldewijk K.K. and Ramankutty N. (2007) Biogeophysical effects of land use on climate: model simulations of radiative forcings and large- scale temporal change. Agricultural and Forest Meteorol. 142, 216-233.

Betts R.A. (1999) Self-beneficial effects of vegetation on climate in an Ocean- Atmosphere General Circulation model. Geophys. Res. Lett. 26, 1457-1460.

Bongers F. and Popma J. (1990) Leaf dynamics of seedlings of rainforest species in relation to canopy gaps. Oecologia 82, 122-127.

Boyce C.K. and Lee J-E. (2010) An exceptional role for flowering plant physiology in the expansion of tropical rainforests and biodiversity. Proc. R. Soc. London, Ser. B 277, 3437-3443.

Buchmann N., Kao W-Y. and Ehleringer J. (1997) Influence of stand structure on carbon- 13 of vegetation, soils, and canopy air within deciduous and evergreen forest in Utah, United States. Oecologia 110, 109-119.

Burnham R.J., Wing S.L. and Parker G.G. (1992) The reflection of deciduous forest communities in leaf litter: Implications for autochthonous litter assemblages from the fossil record. Paleobiol. 18, 30-49.

Burnham R.J. and Johnson K.R. (2004) South American palaeobotany and the origins of neotropical rainforests. Proc. R. Soc. London, Ser. B 359, 1595-1610.

Codron J, Codron D., Lee-Thorp J.A., Sponheimer M., Bond W.J., de Ruiter D. and Grant R. (2005) Taxonomic, anatomical, and spatio-temporal variations in the stable carbon and nitrogen isotopic compositions of plants from an African savanna. J. Archaeolog. Sci. 32, 1757-1722.

96

Coplen T.B., Brand W.A., Gehre M., Groning M., Meijer H.A.J., Toman B. and Verkouteren R.M. (2006) New guidelines for δ13C measurements. Anal. Chem. 78, 2439-2441.

Correa E., Jaramillo C., Manchester S. and Gutierrez M. (2010) A fruit and leaves of Rhamnaceous affinities from the Late Cretaceous (Maastrichtian) of Colombia. Am. J. Bot. 97, 71-79.

Davis C.C., Webb C.O., Wurdack K.J., Jaramillo C.A. and Donoghue M.J. (2005) Explosive radiation of Malpighiales supports a mid-Cretaceous origin of modern tropical rainforests. Am. Naturalist 165, E36-E65.

De la Parra F., Jaramillo C.A. and Dilcher D.L. (2007) Paleoecological changes of spore producing plants through the Cretaceous-Paleocene boundary in Colombia. Palynol.232, 258-259.

Diefendorf A.F., Mueller K.E., Wing S.L., Koch P.L. and Freeman K.H. (2010) Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proc. Natl. Acad. Sci. 107, 5738-5743.

Doria G., Jaramillo C., and Herrera F. (2008) Menispermaceae from the Cerrejón Formation, middle to late Paleocene, Colombia. Am. J. Bot. 95, 954-973.

Ehleringer J.R., Field C.B., Lin Z. and Kuo C. (1986) Leaf carbon isotope and mineral composition in subtropical plants along irradiance cline. Oecologia 70, 520-526.

Ellis B., Daly D.C., Hickey L.J., Johnson K.R., Mitchell J.D., Wilf, P. and Wing S.L. (2009) Manual of Leaf Architecture. Cornell University Press, Ithaca, USA.

Ellsworth D.S. and Reich P.B. (1993) Canopy structure and vertical patterns of photosynthesis and related leaf traits in a deciduous forest. Oecologia 96, 169- 178.

Farley M.B. (1989) Palynological facies fossils in nonmarine environments in the Paleogene of the Bighorn Basin. Palaios 4, 565-573.

Farquhar G.D., Ehleringer J.R. and Hubick K.T. (1989) Carbon isotope discrimination and photosynthesis. Ann. Rev. Physiolog. Plant Mol. Biol. 40, 503-537.

Ferguson D.K (1985) The origin of leaf-assemblages – New light on an old problem. Rev. Palaeobot. Palynol. 46, 117-188.

Feild C.B., Behrenfeld M.J., Randerson J.T. and Falkowski P. (1998) Primary production of the biosphere: integrating terrestrial and oceanic components. Science 281, 237-240.

97

Germeraad J.H., Hopping C.A. and Muller J. (1968) Palynology of Tertiary sediments of tropical areas. Rev. Palaeobot. Palynol. 6, 189-348.

Grace J., Malhi Y., Higuchi N. and Meir P. (2001) Productivity of tropical rainforests. In: Terrestrial global productivity: past, present and future (eds. H. Mooney, J. Roy and B. Saugier). Academic Press, San Diego, USA. pp. 401-426.

Greenwood D.R. (1992) Taphonomic constraints on foliar physiognomic interpretations of Late Cretaceous and Tertiary palaeoclimates. Rev. Palynol. Palaeobot. 71, 149- 190.

Greenwood D.R. (2007) Fossil angiosperm leaves and climate: from Wolfe and Dilcher to Burnham and Wilf. Cour. Forsch.-Inst. Senckenberg 258, 95-108.

Guerrero J. (2002) A Proposal on the Classification of Systems Tracts: Application to the Allostratigraphy and Sequence Stratigraphy of the Cretaceous Colombian Basin. Part 2: Barremian to Maastrichtian. Geología Colombiana 27, 27-49.

Gutierrez M. and Jaramillo C. (2007) Maastrichtian paleotemperature and paleoprecipitation from the Guaduas Formation, Colombia. Palynol. 32, 260.

Hanba Y.T.S., Lei T.T., Koike T. and Wada E. (1997) Variations in leaf δ13C along a vertical profile of irradiance in a temperate Japanese forest. Oecologia 110, 253- 261.

Heaton T.H.E. (1999) Spatial, species, and temporal variations in the 13C/12C ratios of C3 plants: implications for paleodiet studies. J. Archaeolog. Sci. 26, 637-649.

Herrera F., Jaramillo C., Dilcher D., Wing S.L. and Gomez C. (2008) Fossil Araceae from a Paleocene Neotropical rainforest in Colombia. Am. J. Bot. 95, 1659-1683.

Herrera F., Manchester S.R., Hoot S.B., Wefferling K., Carvalho M. and Jaramillo C. (2011) Phytogeographic implications of fossil endocarps of Menispermaceae from the Paleocene of Colombia. Am. J. Bot. 98, 1-14.

IPCC (2000) Land use, land use change and forestry. A special report of the IPCC. Cambridge University Press, Cambridge UK.

Jaramillo C.A., and Dilcher D. (2000) Microfloral diversity patterns of the late Paleocene-Eocene interval in Colombia, northern South America. Geol. 28, 815- 818.

Jaramillo C.A., and Rueda M. (2004) Impact of biostratigraphy on oil exploration. III Convención Técnica Asociación Colombiana de Geólogos y Geófisicos del Petroleo, Bogotá, Colombia.

98

Jaramillo C.A., Pardo-Trujillo A., Rueda M., Harrington G., Bayona G., Torres V. and Mora G. (2007) Palynology of the Upper Paleocene Cerrejón Formation, northern Colombia. Palynol. 31, 153-189.

Jaramillo C.A., Ochoa D., Contreras L., Pagani M., Carvajal-Ortiz H., Pratt L.M., Krishnan S., Cardona, Romero M., Quiroz L., Rodriguez G., Rueda M.J., de la Parra F., Morón S., Green W., Bayona G., Montes C., Quintero O., Ramirez F., Mora G., Schouten S., Bermudez H., Navarrete R., Parra F., Alvarán M., Osorno J., Crowley J.L., Valencia V. and Vervoort J. (2010) Effects of rapid global warming at the Paleocene-Eocene boundary on neotropical vegetation. Science 330, 957-961.

Jaramillo C. and Cárdenas A. (2013) Global warming and neotropical rainforests: A historical perspective. Ann. Rev. Earth Planet. Sci. 41, 741-766.

Johnson K.R. (1992) Leaf-fossil evidence for extensive floral extinction at the Cretaceous-Tertiary boundary, North Dakota, USA. Science 296, 2379-2383.

Johnson K.R. and Ellis B. (2002) A tropical rainforest in Colorado 1.4 million years after the Cretaceous-Tertiary boundary. Cretaceous Res. 13, 91-117.

Kapos V.. Ganade G., Matsui E., and Vicotria R.L. (1993) δ13C as an indicator of edge effects in tropical rainforest reserves. J Ecol. 81, 425-432.

Lang G.E. and Knight D.H. (1983) Tree growth, mortality, recruitment, and canopy gap formation during a 10-year period in a tropical moist forest. Ecol. 64, 1075-1080.

Leidelmeyer P. (1966) The Paleocene and Lower Eocene pollen flora of Guyana. Leidse Geologische Mededelingen 38, 49-70.

Lockheart M.J., Van Bergen P.F. and Evershed R.P. (1997) Variations in the stable carbon isotope compositions of individual lipids from the leaves of modern angiosperms: implications for the study of higher land plant-derived sedimentary organic matter. Org. Geochem. 26, 137-153.

Malhi Y., Meir P. and Brown S. (2002) Forests, carbon and global climate. Philos. Trans. R. Soc. London, Ser. B 354, 1739-1748.

Medina E. and Minchin P. (1980) Stratification of δ13C values of leaves in Amazonian rainforests. Oecologia 45, 377-378.

Meir P., Kruijt B., Broadmeadow M., Barbosa E., Kull O., Carswell F., Nobre A. and Jarvis P.G. (2002) Acclimation of photosynthetic capacity to irradiance in tree canopies in relation to leaf nitrogen concentration and leaf mass per unit area. Plant, Cell and Env. 25, 343-357.

99

Morley R.J. (2000) Origin and Evolution of Tropical Rainforests. John Wiley & Sons, Chichester UK.

Muller J.E., Di Giacomo E. and Van Erve A. (1987) A palynologic zonation for the Cretaceous, Tertiary and Quaternary of northern South America. Am. Ass. of Stratigraphic Palynolgists Cont. Series 19, 7-76.

Myers J.A. and Kitajima K. (2007) Carbohydrate storage enhances seedling shade and stress tolerance in a neotropical forest. J. Ecology 95, 383-395.

Niinemets Ü., Kull O. and Tenhunen J.D. (1999) Variability in leaf morphology and chemical composition as a function of canopy light environment in coexisting deciduous trees. Int. J.Plant Sci. 160, 837-848.

Parrish J.T. (1998) Interpreting pre-Quaternary climate from the geologic record. Columbia University Press, New York.

Reich P.B., Uhl C., Walters M.B. and Ellsworth D.S. (1991) Leaf lifespan as a determinant of leaf structure and function among 23 Amazonian tree species. Oecologia 86, 16-24.

Rice K., Brokaw N., and Thompson J. (2004) Liana abundance in a Puerto Rican forest. Forest Ecology. Management 190, 33-41.

Santiago L.S. and Wright S.J. (2007) Leaf functional traits of tropical forest plants in relation to growth form. Functional Ecol. 21,19-27.

Sarmiento G. (1992) Estratigrafía y medio de deposición de la Formación Guaduas. Boletín Geólog. Ingeominas 32, 3-44.

Spicer R.A. (1981) The sorting and deposition of allochthonous plant material in a modern environment at Sillwood Lake, Sillwood Park, Berkshire, England. USGS Surv. Prof. Paper 1143.

Spicer R.A. and Wolfe J.A. (1987) Plant taphonomy of late Holocene deposits in Trinity (Clair Engle) Lake, northern California. Paleobiol. 13, 227-245.

Spicer R.A. (1989) The formation and interpretation of fossil plant assemblages. Adv. Bot. Res. 16, 95-191.

Spicer R.A., Rees P.M. and Chapman J.L. (1993) Cretaceous phytogeography and climate signals. Proc. R. Soc. London, Ser. B 341, 277-285.

100 Tipple B.J., Meyers S.R. and Pagani M. (2010) Carbon isotope ratio of Cenozoic CO2: A comparative evaluation of available geochemical proxies. Paleoceanography 25, PA302.

Turner I.M. (2001) The Ecology of Trees in the Tropical Rainforest. Cambridge University Press, Cambridge UK.

University of California, San Diego Scripps Global CO2 Program. (2012) Database accessed July 2013, http://scrippsco2.ucsd.edu

Upchurch G.R. and Wolfe J.A. (1987) Mid-Cretaceous to early Eocene vegetation and climate: evidence from fossil leaves and woods. In The Origins of Angiosperms and their Biological Consequences (eds. E.M. Friis, W.G. Chaloner and P.R. Crane). Cambridge University Press, Cambridge UK. pp. 75-105.

Valladares F. and Niinemets Ü., (2008) Shade tolerance, a key plant feature of complex nature and consequences. Ann. Rev. Biol. Ecolog. Evolution, and Systematics 39, 237-257.

Vogel J.C. (1978) Recycling of carbon in a forest environment. Oecologia Plantarum 13, 89–94.

Wilf P., Wing S.L., Greenwood D.R. and Greenwood C.I. (1998) Using fossil leaves as paleoprecipitation indicators: an Eocene example. Geol. 26, 203-206.

Wilf P. and Johnson K.R. (2004) Land plant extinction at the end of the Cretaceous: a quantitative analysis of the North Dakota megathermal record. Paleobiol. 30, 347- 368.

Wilf P., Labandeira C.C., Johnson K.R. and Ellis B. (2006) Decoupled plant and insect diversity after the end-Cretaceous extinction. Science 313, 1112-1115.

Wills C. and Condit R. (1990) Similar non-random processes maintain diversity in two tropical rainforests. Proc. R. Soc. London, Ser. B 266, 1445-1452.

Wing S.L. and Boucher L.D. (1998) Ecological aspects of the Cretaceous flowering plant radiation. Ann. Rev. Earth Planet. Sci. 26, 379-7421.

Wing S.L., Herrera F., Jaramillo C.A., Gómez-Navarro C., Wilf P. and Labandeira C.C. (2009) Late Paleocene fossils from the Cerrejón Formation, Colombia, are the earliest record of Neotropical rainforest. Proc. Natl. Acad. Sci. 106, 18627-18632.

Zimmerman J.K. and Ehleringer J.R. (1990) Carbon isotope ratios are correlated with irradiance levels in the Panamanian orchid Catasetum viridiflavum. Oecologia 83, 247-249.

101 Chapter 4. Plant waxes and fossil leaves from Cretaceous and Paleocene Neotropics: a Comparison of Modern and Ancient Fractionation Factors

4.1 Abstract

The isotopic composition of terrestrial organic matter in the sedimentary record is often used as a proxy for environmental and climatic conditions in the geologic past. This organic matter is presumed to be derived primarily from the leaves of plants and these plants are assumed to have exhibited the same metabolic properties of modern relatives.

A more specific application of isotopic analysis in the geologic record uses molecular fossils from plants, such as n-alkanes from leaf waxes. Alkanes can be used to make more specific observations about sedimentary archives but the environmental, ecologic, and ontogenetic influences on alkane production and carbon isotope expression are very poorly understood.

This study analyzes the carbon isotope composition of fossil leaves and alkanes extracted

13 13 from those fossil leaves (δ Cleaf and δ Clipid) and compares these observations with leaves from the same families in a similar modern biome. 63 fossil leaves were collected from Cretaceous and Paleocene deposits. Cretaceous fossils analyzed are among the oldest angiosperm fossils found in South America and Paleocene fossils are from deposits believed to represent the first true, closed-canopy, Neotropical forest. These fossils are compared with leaves from modern relatives from a forest in Panama. Modern leaves were collected throughout the canopy in order to capture the environmental gradients

102 (light, humidity, CO2) that affect carbon isotope fractionation. This study compares the photosynthetic and lipid biosynthetic fractionation factors in order to understand the nature of carbon metabolism in ancient plants.

4.2 Introduction

The leaves of land plants play an important role as the interface between atmosphere and geosphere. Reconstructions of past terrestrial environments often rely on the carbon isotope composition (δ13C) of preserved organic matter in the sedimentary record that arises from leaves as an indicator of environmental conditions as well as terrestrial community composition (Bowen et al., 2004; Magioncalda et al., 2004; Sauer et al.,

2001; Wing et al., 2005; Schouten et al., 2007; Secord et al., 2008;). Often these interpretations do not fully appreciate the effects that leaf physiognomy and three- dimensional ecosystem structure can have on the plant carbon isotope record. Certain aspects of leaf anatomy (size, vein density, stomatal density) are plastic and can change in response to climate, carbon dioxide concentration, and canopy position (Medina &

Minchin, 1980; Da Silveira et al., 1989; Brooks et al., 1997; Royer, 2001; Beerling &

Royer, 2002; Feild et al., 2011; Franks et al., 2012; Carins et al., 2013). Since leaf morphology affects carbon availability and is reflected in photosynthetic isotope

13 fractionation (∆leaf) these changes should then be considered when using δ COC as an environmental indicator in the sedimentary record (Royer et al.,2005; Brodribb et al.,

2007; Boyce, 2008).

103 The use of organic carbon in geologic archives for the purpose of plant community reconstruction is problematic since terrestrial organic matter can come from many sources besides leaves. For this reason many workers now advocate plant-specific biomarkers, such as n-alkanes, as an better proxy of carbon from ancient terrestrial ecosystems (Bird et al., 1995; Briggs et al., 2000; Lockheart et al., 2000; van Aarssen et al., 2000; Pancost & Boot, 2004; Smith & Freeman, 2006; Smith et al., 2007; Polissar et

13 al., 2009; Jaramillo et al., 2010; Castaneda & Schouten, 2011). δ Clipid can be used to infer environmental or ecologic conditions, however the influence of phylogeny, ecosystem structure, and climate on the carbon isotope fractionation between leaf tissue and lipid (εlipid) is poorly understood (Chikaraishi et al., 2004; Bi et al., 2005;

Pedentchouk et al., 2008; Diefendorf et al., 2011).

This study reports the photosynthetic and lipid biosynthetic fractionation factors for leaf tissue and alkanes from five genera in a modern tropical forest through a vertical transect of forest - from forest floor to top of canopy - following a pronounced humidity and light gradient. These values are compared with a similar analysis of fractionation factors for preserved leaf carbon and extracted alkanes from 63 Cretaceous and Paleocene fossil leaves. These fossil leaves are from a similar tropical climate and many are from the same families as the modern leaves that were analyzed. Fossil leaves are likely from both open and closed forest canopies but were growing in an atmosphere with greater carbon dioxide. Our analyses reveal a wide range in ∆leaf values for modern and fossil leaves that likely reflects the humidity range typical in closed-canopy forests. Calculated εlipid values are highly variable within families and across environmental gradients. Fossil εlipid values

104 were less negative than modern relatives. This may be a reflection of the bias toward canopy leaf preservation in fossil assemblage or may be related to climatic seasonality or dryness in the ancient environment.

4.3. Methods

4.3.1. Collection Sites

Guaduas Formation

The Guaduas flora - from the middle Guaduas Formation of Boyacá department in central

Colombia (see Fig. 1) - remains one of the only sources of plant leaf fossils in the

Maastrichtian of South America. Fossil leaves were collected at 5˚55'45"N, 72˚47’43"W; the paleolatitude of this site during the time of deposition was closer to 2˚N (Sarmiento,

1992; Correa et al., 2010). Although the Guaduas Formation is now 2700 m above sea level in the Andean mountains, at the time of deposition this location was a coastal plain

(Sarmiento, 1992). Reconstruction of mean annual precipitation (MAP) and mean annual temperature (MAT) from leaf size and leaf margin analysis suggest an MAP of ~2400 mm/year and an MAT of 22.1 +/- 3.4˚C (Gutierrez & Jaramillo, 2007). The palynoflora of the assemblage can be assigned to one of two palynological zones that, when calibrated against foraminifera, indicate an age of ca. 68-66 Ma (Muller et al., 1987;

Jaramillo & Rueda, 2004; Correa et al., 2010).

105

Cerrejón Formation, Sites 0315 and 0318

The Cerrejón flora are found in outcrops of the Cerrejón Formation throughout the

Guajira Peninsula of northern Colombia and, most notably, in an open-pit coal mine at the south of the peninsula (see Fig. 1). Samples for this study were taking from two separate beds within the mine located at 11˚5'60"N, 72˚30’0"W (Fig. 2). The estimated paleolatitude of this location was 5˚30'N (Jaramillo et al., 2007). Palynological assemblages collected at the mine indicate a middle-to-late Paleocene age, approximately

60-58 Ma (Bayona et al., 2004; Jaramillo et al., 2007). Leaf margin analysis suggest an

MAT of 24 - 31˚C. Leaf size analysis was used to estimate an MAP of 2,260-4.640 mm/year (Herrera et al., 2008; Wing et al., 2009). These data support other work indicating a warmer, wetter Paleocene when compared to the Cretaceous (Leidelmeyer,

1996; Parrish, 1998; see Burnham & Johnson, 2004 and references therein). The wide variety of sediments (sandstones, mudstones, and coals) in the Cerrejón Formation suggest a mosaic of fluvial and lacustrine settings typical of an estuarine coastal plain

(see Fig. 2) (Jaramillo et al., 2007). Fossils for this study were collected at two separate sites (315 and 318, see Fig. 2) less than 2 km away from one another and vertically 100 m apart. Fossils from site 0315 were found in heterolithic sediments suggestive of a low-to- medium energy channel. Leaves in the overlying site 0138 were found in thinly bedded siltstones interpreted as small lakes.

106 While not as diverse as modern forests, the flora of the Cerrejón include all of the key floristic elements of the extant Neotropical closed-canopy forests. The Cerrejón flora are dominated by Arecaceae, Fabaceae, Lauraceae, Malvaceae, Menispermaceae, Araceae and Zingiberales (Wing et al., 2009). Physiognomic features of leaves from the Cerrejón

– size, entire margins, vein density – also suggest a closed, multistratal rainforest, that is, a closely spaced forest with many small trees, vines, and epiphytes (Wing et al., 2009;

Herrera, 2008; Jaramillo et al., in prep).

San Lorenzo, Panama

The Bosque Protector San Lorenzo forest preserve of Panama (Fig. 1) is located at 9˚

17’N, 79˚38’W. This evergreen, moist tropical terre firme forest is 130 m above sea level and 4.4 km from the Caribbean coast. The site has a tropical monsoonal climate (Köppen classifcation by FAO GeoNetwork) receiving on average 330 cm of rainfall annually and has a mean annual temperature of 26 ºC. The 6 ha preserve contains more than 22,000 trees and 240 recorded species of trees and lianas. The canopy is co-dominated by

Manilkara spp., and Brosimum utile. Aspidosperma cruenta occupy much of the mid- canopy (CTFS; http://www.ctfs.si.edu/site/Sherman). This forest exhibits a marked, year- round light gradient with less than 1% of available light (>99% light attenuation) reaching the forest floor. No logging has occurred on this site for at least 200 years and it has been an officially protected site for over a century.

107 4.3.2. Sampling methods

Fossil cuticle and mesophyll preparation

Cuticle and mesophyll samples from both the Guaduas and Cerrejón floras were selected from fossils stored at the Smithsonian Tropical Research Institute. These fossils had previously been prepared for visual inspection; rocks were split along bedding planes and dusted to reveal leaf structure and morphology. Rock material adjacent to the fossil was removed by scraping or grinding in order to reveal leaf edges. The surfaces of the fossils were kept as clean as possible in order to prevent damage or obfuscation of features. That said, prior to sampling for carbon isotope analysis, fossils were gently brushed and, when necessary, rinsed with deionized water to remove any surface contaminants. Fossil leaves were selected based on preservation and taxonomic association with modern families.

Of the 21 samples taken from the Guaduas flora, only two could be identified to the level of family. Four samples were assigned to three morphotypes, and the other 15 were identified simply as indeterminate dicots. For this study, 23 leaf samples were taken from

Cerrejón site 0135. Specimens sampled included six morphotypes assigned to the families Apocynaceae, Bombaceae, Fabaceae, Lauraceae, Moraceae, and Sapotaceae.

Another 19 leaf samples were taken from Cerrejón site 0318. These samples included ten morphotypes. Eight of these morphotypes have been assigned to the families Fabaceae,

Euphorbaceae, Malvaceae, Meliaceae, Menispermaceae, and Sapotaceae. All

Menispermaceae fossils are allied with the extinct species Menispermites cerrejonensis

108 (Doria et al., 2008). Details of the family identifcation and morphotype assignment of these fossils can be found in Herrera et al. (2008a), Wing et al. (2009), and Correa et al.

(2010).

Cuticle and mesophyll isolates were either picked, scraped, or ground from the rock matrix using either forceps, scalpel or a rotary grinding tool equipped with a diamond- pointed bit, as appropriate. Every effort was made to minimize extraneous material in cuticle samples. Cuticle that could be removed by forceps or scalpel had sediment brushed from surface. Fossils that required vigorous scraping or grinding to remove cuticle or mesophyll were sampled under a dissecting microscope to ensure that a minimum of sediment was introduced into the sample. All samples were crushed and homogenized with a Teflon pestle in glass tubes. Samples were 6N HCl for two hours and rinsed three times with deionized water. This treatment was repeated until the final rinse water maintained a neutral pH. Samples were then lyophilized.

Modern leaf selection and sampling

At the San Lorenzo forest, leaves were collected along a 47-m vertical transect using a canopy crane. This tower crane, managed by the Smithsonian Tropical Research Institute

(STRI), provides access to nearly the entire forest structure in a 0.92 ha area. Leaves were selected from all strata of the forest, from the very canopy top to trees in the understory and in gaps. Bulk leaf samples of Aspidosperma cruenta (Apocynaceae), Brosimum utile

(Moraceae), and Manilkara spp.(Sapotaceae) were taken at 49 unique sampling sites.

109 Leaves were clipped from trees, stored in paper bags, and dried at 70ºC the same day as collection. All samples were acquired during a two-month time window, from January to

February of 2010. After drying, a composite of 20-30 dried leaves from each sampling site was ground and homogenized. More information on collection procedures can be found in chapter 2.

4.3.3. Organic carbon analyses

Carbon isotope and weight percent total organic carbon (wt. % TOC) analyses were performed on ground leaf, fossil, and decarbonated sediment samples using a Costech elemental analyzer connected to a Thermo Finnigan Delta XP IRMS. Analyses were performed under continuous-flow conditions (He; 120 ml/min), oxidized at 1020°C over chromium (III) oxide and cobaltous silver (II, III) oxide, then reduced over elemental copper at 650°C before passing through a water trap and into IRMS. δ13C values were corrected for sample size and reported relative to Vienna Pee Dee Belemnite (VPDB)

(Coplen et al., 2006). Error was determined by repeated analysis of internal standards

(calibrated by IAEA CH-7, IAEA CH-6, and USGS-24; polyethylene, sucrose and graphite, respectively). Instrument precision standard deviation was ±0.09‰ (n=76).

Accuracy, the average difference between the measured and true δ13C value, was 0.02‰

(n = 96). Instrument precision for wt. % TOC measurements was ±1.0% (n=76).

Accuracy, the average difference between the measured and true δ13C value, was 0.97%

(n = 96). Values below 1.0% TOC were not quantifiable. Duplicate analyses were performed for 50% of all samples. Samples that defined the end points of the isotopic

110 range were analyzed in triplicate.

4.3.4. n-Alkane analyses

Fossil Extraction and Preparative Chromatography

Fossil carbon was extracted using a Dionex Accelerated Solvent Extraction system with dichloromethane/methanol (65:35, v/v). Samples were heated for 5 min at 76 bar and

100ºC, followed by a 5-min heated hold, flushed with solvent mix and purged with N2 for

150 s. n-Alkanes were chromatographically separated into polarity fractions by sequential in-cell separation by the Dionex Accelerated Solvent Extraction system as per Magill et al., (in review). Total lipid extracts (TLEs) were then evaporated to near dryness under

N2 before being transferred to an ashed glass fiber disc (Dionex #047017). The TLE- saturated disc was packed in a column with 1 g of high purity silica gel with a a 60 Å pore size and 70-230 mesh particle size and then 1 g of 5% silver-impregnated alumina oxide. Hydrocarbons were eluted in 1:20 dichloromethane:hexanes, and more polar compounds were eluted with ethyl acetate, dichloromethane and methanol. Columns were heated to 50ºC for one minute at 500 psi then flush with 70% of the cell volume three times. The final elution, following the heated hold period, flushed with entire solvent mix and purged the column with N2 for 150 s.

Modern Leaf Extraction and Preparative Chromatography

111 Dried, ground leaf material (0.5 g) was extracted using a Dionex Accelerated Solvent

Extraction system with dichloromethane/methanol (65:35, v/v). Samples were heated for

5 min at 76 bar and 100ºC, followed by a 5 min heated hold, flushed with solvent mix and purged with N2 for 150 s. This process was repeated three times. Total lipid extract were chromatographically separated into polarity fractions with 0.5 g of high purity silica gel with a 60 Å pore size and 70-230 mesh particle size. Hydrocarbons were eluted in

1:20 dichloromethane:hexanes. More polar compounds were eluted with ethyl acetate, dichloromethane and methanol.

Alkane identification

Extracted alkanes were identified using a Hewlett-Packard (HP) 6890 gas chromatograph

(GC) connected to an HP 5973 quadrupole mass spectrometer (MS). We used a fused silica capillary column (Agilent J&W DB-5; 30 m, 0.32 mm, 0.25 µm) with helium carrier gas. Injections were made in hexanes and at an inlet temperature of 320ºC. The column flow rate is 2.0 ml/min and the column oven starts with an initial temperature of

60ºC for 1 min, followed by a ramp to 320ºC at 6ºC/min. This final temperature is then held for 20 min. The MS was operated with a scanning mass range of m/z 50-700 at 3 scans per second and an ionization energy of 70 eV. Compounds were identified using standards, the NIST 98 spectrum library, fragmentation patterns, the published literature, and chromatographic retention times.

Alkane Carbon Isotope Analysis

112

Extracted n-alkane fractions were separated on a Varian model 3400 GC equipped with a split/splitless injector operated in splitless mode with a fused silica column (Agilent J&W

DB-5; 30 m, 0.32 mm, 0.25 µm). The GC temperature program followed the same conditions as the identification and quantification steps. Following separation, compounds were combusted over nickel and platinum wire with O2 in He (1%, v/v) at

1000ºC, with the resulting CO2 monitored using a Finnegan Mat 252. Carbon isotope values of samples are reported in delta notation relative to the VPDB scale. Within-run precision and accuracy was determined by a co-injected an internal standard containing alkanes nC-38 and nC-41. Isotopic abundances were determined relative to a reference gas calibrated with Mix A and Mix B (n-C16 to n-C30; Arndt Schimmelmann, Indiana

University). For all samples, the precision and accuracy were 0.22‰ and 0.06‰ (1σ, n=146), respectively.

4.4. Results

4.4.1 Carbon isotope composition of fossils and leaves

13 The mean δ Cfossil values for both Cerrejón sites were identical, -25.9‰, and similar to the mean for Guaduas (-26.0‰) (see Fig. 3). All fossil values were 13C -enriched relative to San Lorenzo leaves (-29.5‰). Modern leaves were considerably more depleted in 13C than fossil leaves. The enriched composition of fossil leaves compared to modern C3

113 13 plants reflects, in part, the 3.6‰ more enriched δ C composition of CO2 in the ancient atmosphere (Tipple et al., 2010).

13 The range of δ Cleaf values for fossils from all three sites were narrower than the range found in the modern ecosystem at San Lorenzo (Fig. 3) (see Chapter 3). The modern leaves were sampled from the understory to the canopy top. In contrast, fossil assemblages are likely to contain leaves mostly from the upper canopy (see Chapter 2), because upper canopy biomass dominates litter flux and subsequent fossil leaf assemblages (Parker et al., 1999; Osada et al., 2001; see Chapters 3; Graham et al., in review).

13 4.4.2. δ Clipid for Fossils and Leaves

13 δ Clipid values differ by chain length and by family. The magnitude of isotope differences can be large between taxonomic groups and the overall range found in alkanes from modern leaves was nearly 20‰ (Fig. 4). Likewise, the isotope range for alkanes extracted from fossil leaves revealed differences between taxonomic groups but a smaller overall range (Fig. 5).

These patterns were similar for paired fossil and modern taxa although the magnitude of isotope differences in chain lengths differed (See Fig. 6 for modern and ancient examples

13 organized by family). The n-C29 and n-C31 homologues were more depleted in C

114 13 relative to n-C27, n-C33 and n-C35. A comparison of the δ Clipid values for fossil

Sapotaceae leaves assigned to the same morphotype from both Cerrejón localities shows very similar values. Fabaceae leaves of different morphotypes had sightly different

13 patterns of δ Clipid values. The modern analogues for Apocynaceae and Sapotaceae leaves were from trees of a single genus in the upper canopy, the portion of the canopy most likely to be preserved in fossil leaf assemblages (see Chapter 2; Parker et al., 1989;

13 Osada et al., 2001). Modern alkane δ Clipid values for Fabaceae and Malvaceae families were taken from the literature (Rommerskirchen et al., 2006; Vogts et al., 2009).

4.5. Discussion

4.5.1. Photosynthetic fractionation in modern and ancient leaves

Photosynthetic fractionation (∆leaf) is a calculated value that represents the carbon isotope fractionation that occurs during the many metabolic steps between the CO2 that diffuses through stomata and the leaf tissue that is constructed. From Farquhar et al. (1989), ∆leaf is defined as,

∆leaf = (a + (b – a))(ci / ca)

where the constant a is the fractionation during due to diffusion and the constant b is the fractionation during carboxylation reactions. ∆leaf will vary with the available CO2 within a leaf (ci) under conditions of constant concentration of atmospheric CO2 (ca) (Ehleringer

115 & Cerling, 1995) and can also be calculated using the isotopic composition of source CO2 and plant tissue

13 13 13 ∆leaf = (δ Catm - δ Cleaf)/(1 + δ Cleaf / 1000)

The internal leaf CO2 concentration depends on the amount of CO2 entering the leaf through the stomata and the amount of CO2 being removed from the stomata as photosynthate. Water stress leads to a closing of stomata which restricts internal CO2 and leads to decreased ∆leaf values.

Precipitation exerts a strong effect on ∆leaf expression. Leaves from xeric biomes have the lowest observed ∆leaf values while less water-stressed sun leaves of the tropical rain forests have the highest (Diefendorf et al., 2010). Biomes can exhibit wide ranges of ∆leaf values Within the range of observed ∆leaf values in the San Lorenzo forest, understory leaves - which grow in conditions of even higher humidity and lower irradiance (see

Chapter 2; Graham et al., in review) than their sun-lit counterparts - have the most positive ∆leaf values (Fig. 7). While precipitation is the major climate parameter driving stomatal opening the relative humidity at the leaf is a powerful driver of stomatal opening and, by extension, the expression of ∆leaf values.

Aside from the climatic effects on ∆leaf values the anatomy of the leaf itself can also dictate the carbon isotope fractionation during photosynthesis. While the aperture of stomata can change rapidly in response to moisture conditions the number and size of

116 stomata on a leaf can also change between generations of leaves depending on ca.

Stomatal density changes in response to ca are observed in the fossil record (van der

Burgh et al., 1993; van der Water et al., 1994; McElwain & Chaloner, 1995; Royer et al.,

2001; Royer, 2001; Beerling et al., 2002) and have been induced in elevated CO2 experiments (Woodward & Kelly, 1995; Lake et al., 2001; Royer, 2001; Ainsworth &

Rogers, 2007; Carins et al., 2013). Adjustments in the number and size of stomata occur on short-time scales and regulate ci/ca (Franks et al., 2012). If leaves maintain constant ci/ca through stomatal adaptations then CO2 diffusion into the leaf – which is dependent on stomatal opening and moisture balance - is the driving environmental parameter that affect ∆leaf.

In addition to stomatal adaptations, plants have evolved efficient hydraulic systems that increase the water transport to leaves, allowing for cooling and thereby stomatal conductance, gas exchange and photosynthetic rates (Boyce et al., 2010; Brodribb &

Feild, 2010; Feild et al., 2011). Both fossil and modern leaves from gymnosperms and basal angiosperms have much lower vein density than higher, more specialized angiosperms (Taylor et al., 2011). The effects of this innovation in leaf architecture is evident in the very different ∆leaf values of gymnosperms vs. angiosperms (Diefendorf et al., 2011). While vein density must be considered when comparing highly divergent phylogenies (such as gymnosperms and angiosperms) the angiosperm families included in this study originated long before our oldest fossil. The trend towards higher vein density in angiosperms arose ~110-140 mya (Feild et al., 2011) and the gross anatomy of the fossil and modern leaves is largely unchanged.

117

13 13 δ CTOC values for all fossils are offset from modern rainforest δ Cleaf values by ~3.6‰.

13 This is roughly the same as the carbon isotopic difference between modern δ Catm and

13 the ancient δ Catm values reconstructed from benthic foraminifera (Tipple et al., 2010).

This suggests that, if modern and ancient photosynthetic isotope fractionation was the same, no appreciable fractionation of bulk leaf material occurred during fossilization.

13 Reconstructed δ Catm values vary up to 2‰ but this variation is small when compared with naturally occurring variations in leaves of up to of 6-10‰ (see Chapter 2; Graham et al., in review).

Reconstructed CO2 concentrations for the Maastrichtian and Paleocene vary greatly and while the values are debated there was certainly higher pCO2 during these intervals than are found in the modern atmosphere (Royer et al., 2001; Beerling et al., 2009).

Estimations of pCO2 from the stomatal index of fossil leaves from the Paleocene (58-60 mya) are between ~400-600 ppmV and in the range of ~360 and 540 ppmV during the

Maastrichtian (Royer et al., 2001; Beerling et al., 2009; Richey & Upchurch, 2011). pCO2 values based on carbon cycle models suggest atmospheric concentration twice the current amount during the interval our fossils are from (Berner , 1998). Although CO2

13 concentration has been shown to affect δ Cleaf during a single growth season (Schubert

13 & Jahren, 2012) δ Cleaf is generally not considered to have an affect on ∆leaf since, when generating new leaves, plants adjust stomatal numbers in response to pCO2 maintaining constant ci/ca (Lake et al., 2001). Likewise, fossil leaves in our study expressed ∆leaf values consistent with modern relatives despite the difference in pCO2 (see Fig. 7).

118

The effect of relative humidity on ci/ca is evident in the difference between ∆leaf values for leaves from different canopy strata in the modern Panama collection (Fig. 7). The mean

∆leaf value for fossil leaves is more positive than for modern leaves and reflects the bias in paleofloral assemblages towards greater preservation of canopy leaves. At San Lorenzo,

∆leaf values for sun-lit upper canopy leaves are 3.3‰ lower than understory leaves. This decreased discrimination by sun-lit leaves reflects both greater water stress and higher fixation rates than experienced by shaded understory leaves.

4.5.2. Lipid biosynthetic fractionation in modern and ancient leaves

While the fractionation between atmospheric CO2 and leaf tissue is well-characterized, the systematic and environmental sources of fractionation that occur during the many steps between photosynthate and an n-alkane (εlipid) are less well understood (Diefendorf et al., 2011; Magill et al., 2013). Lipids in general have more depleted δ13C compositions when compared to other tissues and metabolites (Hayes, 2001).

Experiments have confirmed an initial negative fractionation occurs when pyruvate is decarboxylated to form Acetyl CoA (DeNiro & Epstein, 1977). Acetyl CoA is the source of acetate in the acetogenic pathway and the precursor to n-alkyl lipids, such as n-alkanes

(Kolattukudy, 1976). Further fractionation occurs during the elongation reactions, when acetate monomers are combined to form alkly chains (Monson & Hayes, 1982). From

Hayes (2001), εlipid is defined as

119 13 13 13 13 εlipid = [(δ Clipid + 1000) / (δ Cleaf + 1000) - 1] x 1000 ≈ (δ Clipid - δ Cleaf)

εlipid is the sum of all carbon isotope fractionation events during synthesis of alkanes,

13 relative to the δ C of bulk leaf tissue. This relationship is often used to derive a value for the δ13C of plant biomass. Alkane proxies allow interpretation of the organic geologic record in absence of plant fossils and/or when sedimentary organic matter contains non-

13 plant material. Often geologic reconstructions use a single εlipid value to interpret δ Clipid values. This practice does not appreciate that in addition to the fractionation constants of

13 synthetic reactions, εlipid is dependent on many other factors such as δ Catm, ci/ca, and phylogenetic, ontogenetic and environmental drivers of lipid synthesis (Rieley et al.,

1991; Lockheart et al., 1997; Chikaraishi et al., 2004; Bi et al., 2005; Pedentchouk et al.,

2008; Diefendorf et al., 2011).

Previous surveys of modern flora have found that εlipid varies between taxanomic groups, especially at the level of plant functional type (Chikaraishi & Naraoka, 2003;

Pedentchouk et al., 2008; Diefendorf et al., 2011). Taxonomic variation in εlipid expression may be due to differing needs for lipid production. More negative εlipid values are associated with greater fractionation during synthesis, which may indicate lower carbon flow to alkane production resulting in a lower amount of alkanes (Hayes, 2001).

Acetate is a precursor in many metabolic reactions and allocation into other pathways will introduce fractionation of the source molecules and affect εlipid (Kollattukudy, 1976;

Hayes, 2001). Indeed, angiosperms produce far greater amounts of alkanes than gymnosperms and express more negative εlipid values as well

120

The seasonal timing of lipid production has also been found to affect εlipid (Pedentchouk et al., 2008). This observation may be related to the large differences in leaf lifespan between angiosperm and gymnosperm plant functional types (Diefendorf et al., 2011).

Recent work suggests that the hydrogen isotope composition of alkanes reflects only the short period of time during leaf flush, implying that an alkanes isotopic composition is set and does not reflect ablation or accumulation changes (Tipple et al., 2013). This would seem to limit the temporal effects on εlipid and may help explain the very wide ranges in

13 δ Clipid values in tropical taxa when compared with temperate taxa (see also Diefendorf et al., 2011). Trees in tropical forests replace leaves nearly constantly throughout the year, whereas temperate forests exhibit a nearly synchronous flush at the beginning of the growing season (see Chapter 2; Graham et al., in review). Widely varying values for

13 δ Clipid and εlipid may be an expression of leaf timing as well as leaf lifespan.

This study considers five taxonomic groups through a light and humidity gradient in a modern closed-canopy forest and compares them with fossil leaves from the same familes and/or climate regimes. In general, εlipid values for fossils have a narrower range than

εlipid values for leaves from San Lorenzo. Although modern leaf εlipid values exhibited a wide range overall, the ranges were narrower within individual families (Fig. 8). The weak association between εlipid values and canopy position may be due to different patterns of lipid production within the canopy. Indeed, modern understory leaves with higher concentrations of alkanes tend to have less negative εlipid values (see Chapter 6). If fossil leaves in this study represent mostly upper canopy leaves, then these results would

121 suggest that the modern canopy leaves in our study produced less alkanes than their

Paleocene relatives.

Alternately, the less negative εlipid values could indicate lower moisture. Vogts et al.

(2009) found that within the same families, εlipid values were higher in savannah specimens when compared to tropical rainforest specimens. The paleoflora in this study are not likely to be savannah communities, but these data might lend caution to paleo- precipitation estimates for these sites, and suggest the possibility that rainfall was more seasonal.

4.6. Conclusions

By comparing modern and fossil leaves from the same families, this study suggests that the amount of CO2 in the atmosphere did not affect the ∆leaf for the taxa and climate conditions represented by these leaves. It appears that ∆leaf has been largely unchanged, at the family level, since the Paleocene. ∆leaf patterns humidity in the modern leaf collection reflects humidity gradients and may account for the wide ranges in the global biome analysis published in Diefendorf et al. (2010).

εlipid values in the modern and fossil leaf collections were highly variable and some values were more negative than previous observations in the literature. Even within individual families εlipid values were highly variable, suggesting strong physiological or environmental controls. In the modern collection, understory leaves tended to be slightly

122 more negative and could be an expression either increased lipid production. Fossil εlipid values were less negative than modern relatives and likely reflects the bias toward canopy leaf preservation in paleofloras or may be related to climatic seasonality or dryness in the ancient environment. In any case, the unanticipated variability suggests that using a single εlipid value to represent the entire angiosperm PFTs may not reflect growth conditions. The nature and controls on εlipid require more investigation if this metric is to be of utility in geologic interpretation.

123 4.7. Figures

Figure 4.1. Modern leaf collection site in Panamá site and three Guaduas and Cerrejón fossil collection sites in two locations in Colombia. Digital elevation model by Amante & Eakins (2009).

124

Figure 4.2. Stratigraphic column of Cerrejón Formation fossil beds from Wing et al., (2009). Fossils were selected from assemblages 0315 and 0318, vertically 100 m apart.

125 13 Figure 4.3. Measured δ Cleaf range for all modern and fossil leaves. The box spans the interquartile range with the median denoted within the box and the whiskers are defined by the upper and lower quartiles.

126

Figure 4.4. Carbon isotope composition for each alkane homologue extracted from leaves of five families of modern trees at San Lorenzo, Panama.

127

Figure 4.5 Carbon isotope composition for each alkane homologue extracted from fossil leaves from the Cerrejón and Guaduas Formations. Data are arranged by morphotype or best guess family as published in Wing et al., 2009. .

128 13 Figure 4.6. Measured δ C25-35 values for alkanes extracted from paired modern and fossil leaves. Plots are arranged by family. Data for modern leaves are plotted in black circles. Fossil data are dark grey and those data for Cerrejón site 0315 are light grey.

129

Figure 4.7. Calculated ∆leaf values for modern leaves from three binned strata of the forest and ∆leaf values for fossil leaves from three study localities. Lower ∆leaf values area associated with upper canopy leaves and all fossil sites express ∆leaf values more typical of upper canopy leaves. .

130

Figure 4.8. Calculated εlipid values for fossil leaves compared with the mean and standard deviation of calculated εlipid values for modern leaves. San Lorenzo leaves are binned by strata. Black lines denote understory while grey denotes canopy. These strata are defined as thos leaf localities below or above 40% available light, respectively. . and ∆leaf values for fossil leaves from three study localities. Higher εlipid values area associated with upper canopy leaves.

131

Figure 4.9. Calculated ∆leaf values for modern leaves from three binned strata of the forest and ∆leaf values for fossil leaves arranged by family from the three study localities. Fossil data are the mean value for all measured individuals or a family and may include multiple morphotypes.

132

Figure 4.10. Total organic carbon measured on all leaves sampled from the Cerrejón and Guaduas Formations. Plots are arranged by locality and include the mean value for measured carbon amounts at each site as well as the mean value for all measured fossil leaves. Data is separated by morphotype to allow comparison of carbon preservation between taxa. Red data represents paired morphotypes from the two Cerrejón sites for comparison of the effects of depostional setting on preservation.

133 4.8. References

Ainsworth E.A. and Rogers A. (2007) The response of photosynthesis and stomatal conductance to rising [CO2]: mechanisms and environmental interactions. Plant Cell Environ. 30, 258- 270.

Bayona G., Jaramillo C.A. Rueda M. Pardo A. and Hernandez G. (2004) Important paleotectonic and chronostratigraphic considerations of the Late Paleocene in the northernmost Andes as constrained by Paleocene rocks in the Cerrejón Coal Mine, Guajira, Colombia. III Convención Técnica Asociación Colombiana de Geólogos y Geófisicos del Petroleo, Bogotá, Colombia.

Beerling D.J. and Royer D.L. (2002) Fossil plants as indicators of the Phanerozoic global carbon cycle. Ann. Rev. Earth Planet. Sci. 30, 527-556.

Berner R.A. (1998) The carbon cycle and CO2 over Phanerozoic time: The role of land plants. Philos. Trans. R. Soc. London, Ser. B 353, 75-82.

Bi, X., Sheng, G., Liu, X., Li, C. and Fu, J. (2005) Molecular and carbon and hydrogen isotopic composition of n-alkanes in plant leaf waxes. Org. Geochem. 36, 1405- 1417.

Bird M.I., Summons R.E., Gagan M.K., Roksandic Z., Dowling L., Head J., Keith Fifield L., Cresswell R.G. and Johnson D.P. (1995) Terrestrial vegetation change inferred from n-alkane δ13C analysis in the marine environment. Geochim. Cosmochim. Acta 59, 2853-2857.

Bowen G.J., Beerling D.J., Koch P.L., Zachos J.C. and Quattlebaum T. (2004) A humid climate state during the Palaeocene/Eocene thermal maximum. Nature 432, 495- 499.

Boyce C.K. (2008) The fossil record of plant physiology and development - what leaves can tell us. Paleontolog. Soc. Papers 14, 133-146.

Boyce C.K., Lee J-E., Field T.S., Brodribb, T.J. and Zwieniecki M.A. (2010) Angiosperms helped put the rain in the rainforests: the impact of plant physiological evolution on tropical biodiversity. Ann. MO Bot. Garden 97, 527- 540.

Briggs D. E. G., Evershed R. P. and Lockheart M. J. (2000) The biomolecular paleontology of continental fossils. Paleobio. 26, 169-193.

Brodribb T.J. and Feild T.S. (2010) An explosion in photosynthetic capacity during early angiosperm diversification. Ecol. Lett. 13, 175-183.

134

Brooks J.R., Flanagan L.B., Varney G.T. and Ehleringer J.R. (1997) Vertical gradients in photosynthetic gas exchange characteristics and refixation of respired CO2 within boreal forest canopies. Tree Physiol. 17, 1-12.

Burnham R.J. and Johnson K.R. (2004) South American palaeobotany and the origins of neotropical rainforests. Proc. R. Soc. London, Ser. B 359, 1595-1610.

Carins Murphy M.R., Jordan G.J. and Brodribb T.J.. (2013) Acclimation to humidity modifies the link between leaf size and the density of veins and stomata. Plant Cell Environ. doi: 10.1111/pce.12136.

Castañeda I.S. and Schouten S. (2011) A review of molecular organic proxies for examining modern and ancient lacustrine environments. Quat. Sci Rev. 30, 2851- 2891.

Chikaraishi, Y., Naraoka, H. and Poulson, S.R. (2004) Hydrogen and carbon isotopic fractionations of lipid biosynthesis among terrestrial (C3, C4 and CAM) and aquatic plants. Phytochem. 65, 1369-1381.

Chikaraishi Y. and Naraoka H. (2003) Compound-specific δD-δ13C analyses of n-alkanes extracted from terrestrial and aquatic plants. Phytochem. 63, 361-371.

Correa E., Jaramillo C., Manchester S. and Gutierrez M. (2010) A fruit and leaves of Rhamnaceous affinities from the Late Cretaceous (Maastrichtian) of Colombia. Am. J. Bot. 97, 71-79.

Da Silveira L., Sternberg L., Mulkey S.S. and Wright S.J. (1989) Oxygen isotope ratio stratification in a tropical moist forest. Oecologia 81, 51-56.

Deniro M.J. and Epstein S. (1977) Mechanism of carbon isotope fractionation associated with lipid synthesis. Science 197, 261-263.

Diefendorf A.F., Mueller K.E., Wing S.L., Koch P.L. and Freeman K.H. (2010) Global patterns in leaf 13C discrimination and implications for studies of past and future climate. Proc. Natl. Acad. Sci. 107, 5738-5743.

Diefendorf A.F., Freeman K.H., Wing S.L. and Graham H.V. (2011) Production of n- alkyl lipids in living plants and implications for the geologic past. Geochim. Cosmochim. Acta 75, 7472-7485.

Ehleringer J.R. and Cerling T.E. (1995) Atmospheric CO2 and the ratio of intercellular to ambient CO2 concentrations in plants. Tree Physiol. 15, 105-111.

Farquhar G.D., Ehleringer J.R. and Hubick K.T. (1989) Carbon isotope discrimination and photosynthesis. Ann. Rev. Plant Physiol. Mol. Bio. 40, 503-537.

135

Feild T.S., Brodribb T.J., Iglesias A., Chatelet D.S., Baresch A., Upchurch G.R., Gomez B., Mohr B.A.R., Coiffard C., Kvacek J. and Jaramillo C. (2011) Fossil evidence for Cretaceous escalation in angiosperm leaf vein evolution. Proc. Natl. Acad. Sci. 108, 8363-8366.

Franks P.J., Letich I.J., Ruszala E.M., Hetherington A.M. and Beerling D.J. (2012) Physiological framework for adaptation of stomata to CO2 from glacial to future concentrations Philos. Trans. R. Soc. London, Ser. B 367, 537-546.

Gutierrez M. and Jaramillo C. (2007) Maastrichtian paleotemperature and paleoprecipitation from the Guaduas Formation, Colombia. Palynology 32, 260.

Hayes, J.M. (2001) Fractionation of carbon and hydrogen isotopes in biosynthetic processes. Rev. Mineral. Geochem. 43, 225-277.

Herrera F., Jaramillo C., Dilcher D., Wing S.L. and Gomez C. (2008) Fossil Araceae from a Paleocene Neotropical rainforest in Colombia. Am. J. Bot. 95, 1659-1683.

Huang Y., Lockheart M.J., Collister J.W. and Eglinton G. (1995) Molecular and isotopic biogeochemistry of the Miocene Clarkia Formation: hydrocarbons and alcohols. Org. Geochem. 23, 785-801.

Jaramillo C.A., and Rueda M. (2004) Impact of biostratigraphy on oil exploration. III Convención Técnica Asociación Colombiana de Geólogos y Geófisicos del Petroleo, Bogotá, Colombia.

Jaramillo C.A., Pardo-Trujillo A., Rueda M., Harrington G., Bayona G., Torres V. and Mora G. (2007) Palynology of the Upper Paleocene Cerrejón Formation, northern Colombia. Palynology 31, 153-189.

Jaramillo C.A., Ochoa D., Contreras L., Pagani M., Carvajal-Ortiz H., Pratt L.M., Krishnan S., Cardona, Romero M., Quiroz L., Rodriguez G., Rueda M.J., de la Parra F., Morón S., Green W., Bayona G., Montes C., Quintero O., Ramirez F., Mora G., Schouten S., Bermudez H., Navarrete R., Parra F., Alvarán M., Osorno J., Crowley J.L., Valencia V. and Vervoort J. (2010) Effects of rapid global warming at the Paleocene-Eocene boundary on neotropical vegetation. Science 330, 957-961.

Kolattukudy P.E. (1976) The Chemistry and Biochemistry of Natural Waxes. Elsevier, Amsterdam.

Lake J.A., Quick W.P., Beerling D.J. and Woodward F.I. (2001) Plant development: signals from mature to new leaves. Nature 411, 154.

136 Leidelmeyer P. (1966) The Paleocene and Lower Eocene pollen flora of Guyana. Leidse Geologische Mededelingen 38, 49-70.

Lockheart M.J., Van Bergen P.F. and Evershed R.P. (2000) Chemotaxonomic classification of fossil leaves from the Miocene Clarkia lake deposit, Idaho, USA based on n-alkyl lipid distributions and principle component analyses. Org. Geochem. 31, 1223-1246.

Magill C.R., Ashley G.M. and Freeman K.H. (2013). Ecosystem variability and early human habitats in eastern Africa. Proc. Nat. Acad. Sci. 4, 1167-1174.

Magioncalda R., Dupuis C., Smith T., Sterubaut E. and Gingerich P.D. (2004) Paleocene- Eocene carbon isotope excursion in organic carbon and pedogenic carbonate: direct comparison in a continental stratigraphic section Geol. 32, 553-556.

McElwain J.C., Beerling D.J. and Woodward F.I. (1999). Fossil plants and global warming at the Triassic-Jurassic Boundary. Science. 285(5432), 1390.

Medina E. and Minchin P. (1980) Stratification of δ13C values of leaves in Amazonian rainforests. Oecologia 45, 377-378.

Monson K.D. and Hayes J.M. (1982) Carbon isotope fractionation in the biosynthesis of bacterial fatty acids. Ozonolysis of unsaturated fatty acids as a means of determining the intramolecular distribution of carbon isotopes. Geochim. Cosmochim. Acta 46, 139-149.

Muller J.E., Di Giacomo E. and Van Erve A. (1987) A palynologic zonation for the Cretaceous, Tertiary and Quaternary of northern South America. Am. Ass. of Stratigraphic Palynolgists Cont. Series 19, 7-76

Osada N., Takeda H., Furukawa A. and Awang M. (2001) Leaf dynamics and maintenance of tree crowns in a Malaysian rain forest stand. J. Ecol. 89, 774-782.

Pancost R.D. and Boot C.S. (2004) The palaeoclimatic utility of terrestrial biomarkers in marine sediments. Marine Chem. 92, 239-261.

Parker G.G., O’Neill J.P. and Higman D. (1989). Vertical profile and canopy organization in a mixed deciduous forest. Vegatatio 85, 1-11.

Parrish J.T. (1998) Interpreting pre-Quaternary climate from the geologic record. Columbia University Press, New York.

Pedentchouk N., Sumner W., Tipple B. and Pagani M. (2008) δ13C and δD compositions of n- alkanes from modern angiosperms and conifers: an experimental set up in central Washington State, USA. Org. Geochem. 39, 1066-1071.

137 Peters-Kottig W., Strauss H. and Kerp H. (2006) The land plant δ13C record and plant evolution in the Late Palaeozoic. Palaeogeog. Palaeoclimatol. Palaeoecol. 240(1- 2), 237-252.

Polissar P., Freeman K.H., Rowley D., McInerney F.A. and Currie B., (2009) Paleoaltimetry of the Tibetean Plateau from D/H ratios of lipid biomarkers. Earth Planet. Sci. Lett. 287, 64-76.

Richey J.D. and Upchurch G.R. (2011) Inference of pCO2 levels during the Late Cretaceous using fossil Lauraceae. American Geophysical Union Fall Meeting 2011, Abstract #PP11A-1771, San Francisco, USA.

Rieley G., Collier R.J., Jones D.M., Eglinton G., Eakin P.A. and Fallick A.E. (1991) Sources of sedimentary lipids deduced from stable carbon isotope analyses of individual n-alkanes. Nature 352, 425-427.

Rommerskirchen R., Plader A., Eglinton G., Chikaraishi Y. and Rullkötter J. (2006) Chemotaxonomic significance of distribution and stable carbon isotopic composition of long-chain alkanes and alkan-1-ols in C4 grass waxes. Org. Geochem. 37, 1303-1332.

Royer D.L., Wing S.L., Beerling D.L., Jolley D.W., Koch P.L., Hickey L.J. and Berner R.A. (2001) Paleobotanical evidence for near present-day levels of atmospheric CO2 during part of the Tertiary. Science 292, 2310-2313.

Royer D.L. (2001) Stomatal density and stomatal index as indicators of paleoatmospheric CO2 concentration. Rev. Palaeobot. Palynol. 114, 1-28.

Royer D.L., Wilf P., Janesko D.A., Kowalski E.A. and Dilcher D.L. (2005) Correlating climate and plant ecology to leaf size and shape: potential proxies for the fossil record. Am. J. Bot. 92, 1141-1151.

Sarmiento G. (1992) Estratigrafía y medio de deposición de la Formación Guaduas. Boletín Geólog. Ingeominas 32, 3-44.

Sauer P.E., Eglinton T., Hayes J.M., Scimmelmann A. and Sessions A. (2001) Compound-specific D/H ratios of lipid biomarkers from sediments as a proxy for environmental and climatic conditions. Geochim. Cosmochim. Acta 65, 213-222.

Schouten S., Woltering M., Rijpstra W.I.C., Slujis A., Brinkhuis H. and Sinninghe Damsté J.S. (2007). The Paleocene-Eocene carbon isotope excursion in higher plant organic matter: Differential fractionation of angiosperms and conifers in the Arctic. Earth Planet. Sci. Lett. 258(3-4), 581-592.

138 Schubert B.A. and Jahren A.H. (2012) The effect of atmospheric CO2 concentration on carbon isotope fractionation in C3 land plants Geochim. Cosmochim. Acta 96, 29- 43.

Secord R., Wing S.L. and Chew A. (2008) Stable isotopes in early Eocene mammals as indicators of forest canopy structure and resource partitioning. Paleobiol. 34,282- 300.

Smith F.A. and Freeman K.H. (2006) Influence of physiology and climate on δD of leaf wax n- alkanes from C3 and C4 grasses. Geochim. Cosmochim. Acta 70, 1172- 1187.

Smith F.A., Wing S.L. and Freeman K.H. (2007) Carbon and hydrogen isotope compositions of plant lipids during the PETM as evidence for the response of terrestrial ecosystems to rapid climate change. Earth Planet Sci. Lett. 262, 50-65.

Tipple B.J., Meyers S.R. and Pagani M. (2010) Carbon isotope ratio of Cenozoic CO2: A comparative evaluation of available geochemical proxies. Paleoceanography 25, PA302. van Aarssen B.G.K., Alexander R. and Kagi R.I. (2000) Higher plant biomarkers reflect palaeovegetation changes during Jurassic times. Geochim. Cosmochim Acta. 64, 1417- 1424. van der Burgh J., Visscher H., Dilcher D.L. and Kürschner W.M. (1993) Paleoatmospheric signatures in Neogene fossil leaves. Science 260, 1788-1790. van der Water P.W., Leavitt S.W. and Betancourt J.L. (1994) Trends in stomatal density and 13C/12C ratios of Pinus flexilis needles during last glacial-interglacial cycle. Science 264, 239-243.

Vogts A., Moossen H., Rommerskirchen F. and Rullkötter J. (2009) Distribution patterns and stable carbon isotopic composition of alkanes and alkan-1-ols from plant waxes of African rain forest and savanna C3 species. Org. Geochem. 40, 1037- 1054.

Wing S.L., Herrera F., Jaramillo C.A., Gómez-Navarro C., Wilf P. and Labandeira C.C. (2009) Late Paleocene fossils from the Cerrejón Formation, Colombia, are the earliest record of Neotropical rainforest. Proc. Natl. Acad. Sci. U.S.A.106, 18627- 18632.

Woodward F.I. and Kelly C.K. (1995) The influence of CO2 concentration on stomatal density. New Phytologist 131, 311-327.

139 Chapter 5. Simulated and observed alkane distribution and isotope composition in sediments using biogeochemical analysis of fossil leaves

5.1 Abstract

The relative chain-length distribution and carbon-isotope composition of n-alkanes extracted from sedimentary rocks are common tools for interpreting properties of terrestrial ecosystems in the absence of macrofossil remains. Alkanes preserved in the rocks are assumed to be contemporaneous, derived from the same ecosystem and integrated from the biomass present on the landscape at that time, yet alkane production can be greatly influenced by the phylogenetic affiliation of taxa as well as environmental parameters and elements of forest structure. Interpretations of these archives would be strengthened by a better understanding of determine how preserved alkanes represent the ecosystem from which they arise and the taphonomic pressures that influence the lipid record.

In this study, we analyze the alkane chain-length distribution and carbon-isotope composition of alkanes extracted from a selection of fossil leaves from two paleofloras.

We then compare a biomass-weighted mixing model based on these data to represent the influence of community composition and forest structure. The model results are compared with alkane measurements from sedimentary matrices collected at each of the assemblage sites. Alkanes extracted from bulk rock samples do show consistencies with the model, a biomass-weighted, integration of alkanes from the major taxa represented in

140 the macrofossil record. Minor differences in observed and modeled chain-length distribution and isotope expression may be due to the biases inherent in the fossil record and an incomplete representation of all taxa.

5.2 Introduction

High-molecular weight (C25 – C35), odd-carbon-numbered n-alkane homologues originate in leaf waxes and are replete in the rock record, often independent of plant macrofossil remains (Eglinton & Hamilton, 1963; Kolattukudy, 1976; Cranwell, 1981). Relative to leaf biomass, cuticular n-alkanes are resistant to chemical change and microbial alteration (Eglinton & Logan, 1991; van Bergen et al., 1995). The molecular distribution

13 and carbon isotope composition of sedimentary alkanes (δ Clipid) have been widely used to reconstruct environmental and climatic conditions and predict dominant plant functional types in ancient ecosystems, in the absence of plant macrofossils (Bird et al.,

1995; Freeman & Colarusso, 2001; Smith et al., 2007; Schouten et al., 2007; Tipple &

Pagani, 2010; Magill et al., 2013). In these applications, alkanes in sediments are commonly considered to represent a simple integration of plant biomass on a landscape.

These interpretations often do not account for the environmental, ecologic, phylogenetic, and ontogenetic influences on alkane production and isotope fractionation between leaves and lipid (εlipid).

Taxonomic groups have differing εlipid patterns and produce characteristic amounts of alkanes and possibly distinctive distributions of alkanes (Chikaraishi & Naraoka, 2003;

141 Rommerskirchen et al., 2006; Vogts et al., 2009). Recent work has shown that three- dimensional forest structure and certain phylogenetic groups can have disproportionate influences on the sedimentary alkane record (Diefendorf et al., 2010; Magill et al., 2013;

Henderson et al., in review; also Chapter 3). Reconstruction of a paleoenvironment from sedimentary n-alkanes requires an understanding of the controls on production, distribution, and flux of lipid biomarkers within ecosystems and the influence of taxa, ecosystem structure, and climate conditions on lipid expression. If the taxa and lipid properties of fossil leaves are representative of plants in an ancient ecosystem, then a leaf assemblage-weighted integration of alkanes preserved in fossil leaves has the potential to

13 reproduce the observed molecular and δ Clipid characteristics in the bulk rock. Analysis of lipids preserved in leaf fossils compared to the alkanes preserved in the ancient sediments of a paleoflora are used here to understand how taxonomic and litter flux characteristics can influence alkanes in sediments.

Relative concentrations of extractable alkane homologs from fossil leaves, and their

13 δ Clipid values were weighted by the contribution of a taxonomic group to the assemblage as determined by a census of fossil leaves. Our approach accounts for both taxonomic distribution and the greater flux from upper canopy leaves estimated by fossil leaf bulk

13 δ Cleaf values (Parker et al., 1999; Osada et al., 2001; see Chapter 3; Graham et al., in review). Our study considered two related Paleocene fossil deposits, one that is believed to be the first closed-canopy forests in the Neotropics and the second that likely represents a forest margin (Wing et al., 2009; Graham et al., in review). The weighted

13 and integrated alkane distributions and δ Clipid compositions were compared to the

142 13 distributions and δ Clipid compositions of alkanes extracted from the matrix of outcrop rock samples from the same sites.

5.3. Methods

5.3.1. Collection Sites

Cerrejón Formation, Sites 0315 and 0318

The Cerrejón flora crops out of the Cerrejón Formation throughout the Guajira Peninsula of northern Colombia and, most notably, in an open-pit coal mine at the south of the peninsula (see Fig. 1). Samples for this study were taking from two separate beds within the mine located at 11˚5'60"N, 72˚30’0"W (Fig. 2). The estimated paleolatitude of this location was 5˚30'N (Jaramillo et al., 2007). Palynological assemblages collected at the mine indicate a middle-to-late Paleocene age, approximately 60-58 Ma (Bayona et al.,

2004; Jaramillo et al., 2007). Leaf margin analysis suggest an MAT of 24 - 31˚C. Leaf size analysis was used to estimate an MAP of 2,260-4.640 mm/year (Herrera et al.,

2008a; Wing et al., 2009). These data support other work indicating a warmer, wetter

Paleocene when compared to the Cretaceous (Leidelmeyer, 1996; Parrish, 1998; see

Burnham & Johnson, 2004 and references therein). The wide variety of sediments

(sandstones, mudstones, and coals) in the Cerrejón Formation suggest a mosaic of fluvial and lacustrine settings typical of an estuarine coastal plain (see Fig. 2) (Jaramillo et al.,

2007). Fossils for this study were collected at two separate sites (315 and 318, see Fig. 2)

143 less than 2 km away from one another and vertically 100 m apart. Fossils from site 0315 were found in heterolithic sediments suggestive of a low-to-medium energy channel.

Leaves in the overlying site 0138 were found in thinly bedded siltstones interpreted as small lakes.

While not as diverse as modern forests, the flora of the Cerrejón include all of the key floristic elements of closed-canopy forests found in the Neotropics today. The Cerrejón flora are dominated by Arecaceae, Fabaceae, Lauraceae, Malvaceae, Menispermaceae,

Araceae and Zingiberales (Wing et al., 2009). Physiognomic features of leaves from the

Cerrejón – size, entire margins, vein density – also suggest a closed, multistratal rainforest, that is, a closely spaced forest with many small trees, vines, and epiphytes

(Wing et al., 2009; Herrera, 2008a; Jaramillo et al., in prep).

5.3.2. Fossil organic material and bulk rock sampling and preparation

Compression fossils from both localities have preserved carbonized leaf material (i.e. phytoleim, which are the remains of cuticle and mesophyll). Samples from both Cerrejón floras were selected from fossils stored at the Smithsonian Tropical Research Institute.

These fossils had previously been prepared for visual inspection; rocks were split along bedding planes and dusted to reveal leaf structure and morphology. Rock material adjacent to the fossil was removed by scraping or grinding in order to reveal leaf edges.

The surfaces of the fossils were kept as clean as possible in order to prevent damage or obfuscation of features. That said, prior to sampling for carbon isotope analysis, fossils

144 were gently brushed and, when necessary, rinsed with deionized water to remove any surface contaminants. Fossil leaves were selected based on preservation and taxonomic association with modern families.

For this study, 23 leaf samples were taken from Cerrejón site 0135. Specimens sampled included six morphotypes assigned to the families Apocynaceae, Bombaceae, Fabaceae,

Lauraceae, Moraceae, and Sapotaceae. Another 19 leaf samples were taken from

Cerrejón site 0318. These samples included ten morphotypes. Eight of these morphotypes have been assigned to the families Fabaceae, Euphorbaceae, Malvaceae, Meliaceae,

Menispermaceae, and Sapotaceae. All Menispermaceae fossils are allied with the extinct species Menispermites cerrejonensis (Doria et al., 2008). Details of the family identifcation and morphotype assignment of these fossils can be found in Doria et al.,

(2008) and Wing et al. (2009).

Fossil organic material was either picked, scraped, or ground from the rock matrix using either forceps, scalpel or a rotary grinding tool equipped with a diamond-pointed bit, as appropriate. Every effort was made to minimize extraneous material in cuticle samples.

Cuticle that could be removed by forceps or scalpel had sediment brushed from surface.

Fossils that required vigorous scraping or grinding to remove cuticle or phytoleim were sampled under a dissecting microscope to ensure that a minimum of sediment was introduced into the sample. All samples were crushed and homogenized with a Teflon pestle in glass tubes then treated with 6N HCl for two hours and rinsed three times with

145 deionized water. This treatment was repeated until the final rinse water maintained a neutral pH. Samples were then lyophilized.

Bulk rock samples were milled to a fine powder, decarbonized using the same protocol described above, and lyophilized prior to isotopic analysis. Bulk rock samples were chipped from the same blocks as fossil leaves but did not contain any visible fossil leaves or leaf fragments.

5.3.3. Organic carbon analyses

Carbon isotope and weight percent total organic carbon (wt. % TOC) analyses were performed on ground leaf, fossil, and decarbonated sediment samples using a Costech elemental analyzer connected to a Thermo Finnigan Delta XP IRMS. Analyses were performed under continuous-flow conditions (He; 120 ml/min), oxidized at 1020°C over chromium (III) oxide and cobaltous silver (II, III) oxide, then reduced over elemental copper at 650°C before passing through a water trap and into IRMS. δ13C values were corrected for sample size and reported relative to Vienna Pee Dee Belemnite (VPDB)

(Coplen et al., 2006). Error was determined by repeated analysis of internal standards

(calibrated by IAEA CH-7, IAEA CH-6, and USGS-24; polyethylene, sucrose and graphite, respectively). Instrument precision standard deviation was ±0.09‰ (n=76).

Accuracy, the average difference between the measured and true δ13C value, was 0.02‰

(n = 96). Instrument precision for wt. % TOC measurements was ±1.0% (n=76).

Accuracy, the average difference between the measured and true δ13C value, was 0.97%

146 (n = 96). Values below 1.0% TOC were not quantifiable. Duplicate analyses were performed for 50% of all samples. Samples that defined the end points of the isotopic range were analyzed in triplicate.

5.3.4. n-Alkane analyses

Alkane extraction and chromatography

Fossil material and associated rock samples were extracted using a Dionex Accelerated

Solvent Extraction system with dichloromethane/methanol (65:35, v/v). Samples were heated for 5 min at 76 bar and 100ºC, followed by a 5-min heated hold, flushed with solvent mix and purged with N2 for 150 s. n-Alkanes were chromatographically separated into polarity fractions by sequential in-cell separation by the Dionex Accelerated Solvent

Extraction system as per Magill et al., (in prep). Total lipid extracts (TLEs) were then evaporated to near dryness under N2 before being transferred to an ashed glass fiber disk

(Dionex #047017). The TLE-saturated disc was packed in a column with 1 g of high purity silica gel with a a 60 Å pore size and 70-230 mesh particle size and then 1 g of 5% silver-impregnated alumina oxide. Hydrocarbons were eluted in a mixture of 1:20 dichloromethane:hexanes, and more polar compounds were eluted with ethyl acetate, dichloromethane and methanol. Columns were heated to 50ºC for one minute at 500 psi, then flushed with 70% of the cell volume three times. The final elution, following the heated hold period, flushed columns with entire the solvent mix and purged the column with N2 for 150 s.

147

Alkane identification and quantification

Extracted alkanes were identified using a Hewlett-Packard (HP) 6890 gas chromatograph

(GC) connected to an HP 5973 quadrupole mass spectrometer (MS). We used a fused silica capillary column (Agilent J&W DB-5; 30 m, 0.32 mm, 0.25 µm) with helium carrier gas. Injections were made in hexanes and at an inlet temperature of 320ºC. The column flow rate is 2.0 ml/min and the column oven starts with an initial temperature of

60ºC for 1 min, followed by a ramp to 320ºC at 6ºC/min. This final temperature is then held for 20 min. The MS was operated with a scanning mass range of m/z 50-700 at 3 scans per second and an ionization energy of 70 eV. Compounds were identified using standards, the NIST 98 spectrum library, fragmentation patterns, the published literature, and chromatographic retention times.

Compounds were quantified on a HP 5890 GC with a flame ionization detector (FID) under the same column and GC conditions as for identification. Compound peak areas were normalized to co-injected internal standard containing alkane n-C16 and 1,1’- binaphthyl. Aliquots were converted to mass quantities using response curves for 28 n- alkanes ranging from n-C12 to n-C40 analyzed in concentrations ranging from 0.2 to 200

µg/ml. Standards were used to determine the accuracy (5.3%; 1σ, n=90) and precision

(4.2%; 1σ, n= 90) of measurements. Concentrations were normalized to the mass of total organic carbon in dry leaf, fossil or rock material extracted.

148 Compound-specific carbon isotope analysis

Extracted n-alkane fractions were separated on a Varian model 3400 GC equipped with a split/splitless injector operated in splitless mode with a fused silica column (Agilent J&W

DB-5; 30 m, 0.32 mm, 0.25 µm). The GC temperature program followed the same conditions as the identification and quantification steps. Following separation, compounds were combusted over nickel and platinum wire with O2 in He (1%, v/v) at

1000ºC, with the resulting CO2 monitored using a Finnegan Mat 252. Carbon isotope values of samples are reported in delta notation relative to the VPDB scale. Within-run precision and accuracy was determined by a co-injected an internal standard containing alkanes nC-38 and nC-41. Isotopic abundances were determined relative to a reference gas calibrated with Mix A and Mix B (n-C16 to n-C30; Arndt Schimmelmann, Indiana

University). For all samples, the precision and accuracy were 0.22‰ and 0.06‰ (1σ, n=146), respectively.

5.3.5. n-Alkane mixing model

Using the relative abundance of leaves reported in a published census of each leaf assemblage (Wing et al., 2009) and the concentration of alkanes in each fossil morphotype group (herein referred to as f) we used an leaf abundance balanced mixing model

13 13 13 13 13 13 13 δ CM = f25δ C25 + f27δ C27 + f29δ C29 + f31δ C31 + f33δ C33 + f35δ C35

149 where f25 + f27 + f29 + f31 + f33 + f35 = 1

13 to predict the δ Clipid for each alkane homologue that would likely be preserved in the bulk rock (see Philips & Koch, 2001). The value f represents the relative percent contribution of a morphotype to the assemblage as well as the average relative yields for each alkane chain expressed by that morphotype (see Appendix from Wing et al., (2009) for morphotype census data).

We also compute the relative lipid abundance and calculate the ACL29-35 for bulk rock.

ACL29-35 = ((f29 *29) + (f31*31) + (f33 *33) + (f35 *35))/(f29+ f31 + f33 + f35)

The average chain length (ACL) of alkanes represented in an assemblage is a representation of relative input of all chain-lengths into the sedimentary archive.

5.4. Results

5.4.1 n-Alkane yields

Of the 42 Cerrejón leaves extracted, 27 had measurable amounts of alkanes. Of the 15 leaves that did not yield enough alkanes to be quantified, 13 had trace quantities of n- alkanes while two had no detectable lipids of any kind. High temperature extraction at

150 150°C was attempted on these samples, as well as increased pressure and solvent hold times, but these efforts did not result in increased yields of identifiable alkanes.

The 27 fossils with quantifiable plant-wax alkanes had, on average, 3.7 mg of alkanes per gram of organic carbon (mg/g OC) – which is roughly 2/3 the average yield for fossils from the well-preserved Miocene-aged Clarkia formation (Lockheart et al., 2000). The amount of alkanes extracted from fossils was highly variable. The amount of alkanes in fossils from Cerrejón site 0315 was 3.2 ± 1.5 mg/g OC and fossils from Cerrejón site

0318 had 4.2 ± 1.4 mg/g OC. Bulk rock samples taken from the Cerrejón Formation had average alkane concentrations of 4.4 mg/g OC.

13 5.4.2 δ Clipid for fossils and sediments

13 δ Clipid values differed by chain length and these patterns were similar for paired fossil and modern taxa although the magnitude of isotope differences in chain lengths differed.

13 n-C29 was depleted in C relative to n-C27, n-C33 and n-C35 (Fig. 3). Similar patterns were

13 found in the δ Clipid of alkanes extracted from bulk outcrop rock. The isotopic difference between shortest and longest alkane homologues extracted from fossil leaves can be up to

~5‰ but only ~1.5‰ for alkanes from bulk rock.

5.5. Discussion

151 5.5.1. n-Alkane yield from fossil leaves

For the purposes of this study, all alkane yields are reported in relative amounts. Not all leaves preserved measurable amounts of alkanes and the extracted amount was highly variable between family groups and between assemblages. In modern leaf collections, alkane concentrations can differ widely between and within taxonomic groups (see

Chapter 6 and Henderson et al., in review). This makes it difficult to judge if the amount of alkanes preserved in a fossil leaf correlates to the amount of alkanes that were produced in the living leaf or if there has been post-depositional loss of alkane content.

There Fossil leaves from the same families and morphotypes have dissimilar yields between sites but yields are similar for morphotypes in the same site. Although this could indicate that alkane concentration is dependent on preservation environment, we found fossils with very little and highly abundant alkanes within the same assemblages. Further, in this study, between fossil leaf %TOC and alkane concentrations for any of the paleofloras.

5.5.2. Alkane chain-length distribution patterns in fossil leaves

Average n-alkane chain lengths are calculated using abundances with the following formula modified from Eglinton & Hamilton (1967) by Henderson et al. (in review),

ACL29-35 = ((C29*29) + (C31*31) + (C33 *33) + (C35 *35))/(C29 + C31 + C33 + C35)

152 Where C29 - C35 is the measured abundance of each alkane homologue. The calculated

ACL29-35 values for fossil leaves and modern leaf values taken from the literature

(Rommerskirchen et al., 2006; Vogts et al., 2009; also see Chapter 4) overlapped for all six families studied (Apocynaceae, Euphorbiaceae, Fabaceae, Lauraceae, Malvaceae, and

Sapotaceae) (Fig. 4). The ACL29-35 for fossils was not statistically different by taxa or from modern leaves. That is, a fossil Fabaceae leaf does not have a statistically unique

ACL29-35 when compared with the ACL29-35 of the five other fossil families. Neither does a modern Fabaceae leaf have an ACL29-35 that would distinguish it from any other modern leaf from those five families (p values range from 0.124 to 0.916). While a fossil leaf’s ACL29-35 does correspond to it’s modern relative neither fossil nor modern leaves are distinguishable from any other taxonomic group by this metric. A compilation of published ACL29-35 values for 205 angiosperm individuals from 56 families, 28 orders, and 9 clades also indicates that ACL29-35 is not a useful proxy for identifying woody C3 vegetation (Bush & McInerney, 2013; Henderson et al., in review).

5.5.3. n-Alkanes from sediments integrate n-alkane patterns from leaves

While the amounts are smaller, the alkanes in bulk rock material have narrower isotopic ranges than fossil leaves, less than 3‰ for all sites. In contrast, the mean isotopic range in

13 13 13 13 alkanes (most C enriched homologue δ Clipid – most C depleted homologue δ Clipid) from a single fossil leaf is ~7‰ , with some morphotypes expressing ranges up to 10‰.

For comparison, individual modern leaves sampled at San Lorenzo have alkane isotope ranges approaching 10‰ (see Chapter 4).

153

Using the relative abundance of leaves in each assemblage as well as the relative amounts of alkanes in each fossil morphotype group, we used the basic mixing model

13 13 13 13 δ CM = fxδ Cx + fyδ Cy + fzδ Cz where fx + fy + fz = 1

We use published census data from Wing et al. (2009) and alkane abundances for each homologue to modify the carbon isotope composition of each homologue

13 13 13 13 13 13 13 δ CM = f25δ C25 + f27δ C27 + f29δ C29 + f31δ C31 + f33δ C33 + f35δ C35

where f25 + f27 + f29 + f31 + f33 + f35 = 1

and f = [alkane] * % morphotype contribution in assemblage

13 to predict the ACL29-35 and δ Clipid that would likely be preserved in the bulk rock (see

Philips & Koch, 2001). Uncertainty of modeled data is represented by the propagation of error from each measured property.

This model considers the relative biomass flux from each family as indicated in the census of leaves (Wing et al., 2009). This model determined the relative influence each

13 taxon had on the chain-length and δ Clipid values of the integrated pool delivered by litter to the soil or sediment.

154 Using the leaf census data for each Cerrejón site we find that the taxa selected for this study represent 73% of the leaves from site 0315 but only 48% of leaves found at site

0318 (Wing et al., 2009). Other major taxa at Cerrejón site 0318 that are not included are

Aracaceae (12.8% of fossil leaves) and Violaceae (10.8% of fossil leaves). The model estimates (Fig. 5) for ACL29-35 at both Cerrejón sites were higher (30.1 for both) but within one standard deviation of the ACL29-35 for measured bulk rock ((29.0 and 28.3, respectively; F(1,6) = 0.498, p = 0.553). Measured alkane components in the fossil leaves from Cerrejón site 0315 had an overall isotopic range of 7.7‰ while from site 0318 values ranged 8.3‰. Integrated isotope values for alkanes calculated by the model ranged

3‰ for site 0315 and 4.4% for site 0318. Even though only 48% of flora at Cerrejón site

0318 were included in the model, the integrated model results matched within 0.5‰ the alkane range found in bulk rock from the site. This likely reflects the sampling emphasis on major floristic elements rather than trying to encompass the diversity of flora at the site.

13 The isotopic composition of modeled δ Clipid values is ~1.1‰ ± 0.09‰ more enriched in

13 C than measured alkanes, and of all homologues. Model estimates for n-C25 and n-C27 were least like their observed values. The ~1.0‰ uncertainty in the model is determined

13 by the propagation of error in the alkane abundance data and δ Clipid values. The model included 44% and 72% of the morphotypes represented in the assemblages, and potentially, this difference could represent alkane input not captured in the mixing model.

155 5.6. Conclusion

13 An analysis of the distribution and δ Clipid of alkanes extracted from fossil leaves as well as the sediments from the fossil assemblage allow us to determine the quantitative link between the leaves in the deposit and the lipid profile in the rock. A biomass weighted mixing model that combined fossil leaf census data from the Cerrejón sites (Wing et al.,

13 2009) with δ Cipid data collected for all alkane homologues produced in fossil leaves predicted alkane profiles and an isotopic range that are similar to those observed in bulk rock. The relative amounts of alkanes modeled were also nearly identical to the ACL29-35 observed in rock samples however a comparison of ACL29-35 values from fossil leaves compared with ACL29-35 values from leaves in modern families found that the ranges were the same in modern and ancient leaves but were not diagnostic. There was no distinctive difference between families. This suggests ACL29-35 is not a valid metric for distinguishing types of woody vegetation. The modeled data nearly matched the observed

(within 1.1‰) remarkably even though analyses lacked data from all morphotypes present in the fossil assemblage.

156 5.7. Figures

Figure 5.1. Modern leaf collection site in Panamá site and three Guaduas and Cerrejón fossil collection sites in two locations in Colombia. Digital elevation model by Amante & Eakins (2009).

157

Figure 5.2. Stratigraphic column of Cerrejón Formation fossil beds from Wing et al., (2009). Fossils were selected from assemblages 0315 and 0318, vertically 100 m apart.

158

Figure 5.3. Average chain length for all measured morphotypes from the Cerrejón Formation grouped by family. ACL data is paired with similar data from modern representatives of these families. Modern plant data is either from the San Lorenzo study (see Chapter 4) or taken from the literature.

159

Figure 5.4. Carbon isotope composition for each alkane homologue extracted from fossil leaves from the Cerrejón Formation. Data are arranged by morphotype or best guess family as published in Wing et al., 2009.

160

13 Figure 5.5. Measured δ C25-35 values for alkanes extracted from bulk rock paired with 13 modeled δ C25-35 values. Bars represent relative amount of alkanes. Modeled data are dark grey and data collected from rock samples are blue.

161 5.8. References

Bird M.I., Summons R.E., Gagan M.K., Roksandic Z., Dowling L., Head J., Keith Fifield L., Cresswell R.G. and Johnson D.P. (1995) Terrestrial vegetation change inferred from n-alkane δ13C analysis in the marine environment. Geochim. Cosmochim. Acta 59, 2853-2857.

Burnham R.J. and Johnson K.R. (2004) South American palaeobotany and the origins of neotropical rainforests. Proc. R. Soc. London, Ser. B 359, 1595-1610.

Bush R.T. and McInerney F.A. (2013) Leaf wax n-alkane distributions in and across modern plants: implications for paleoecology and chemotaxonomy. Geochim. Cosmochim. Acta 117, 161-179.

Chikaraishi Y. and Naraoka H. (2003) Compound-specific δD-δ13C analyses of n-alkanes extracted from terrestrial and aquatic plants. Phytochem. 63, 361-371.

Cranwell P.A. (1981) Diagenesis of free and bound lipids in terrestrial detritus deposited in a lacustrine sediment. Org. Geochem. 3(3), 79-89.

Diefendorf A.F., Freeman K.H., Wing S.L. and Graham H.V. (2011) Production of n- alkyl lipids in living plants and implications for the geologic past. Geochim. Cosmochim. Acta 75, 7472-7485.

Doria G., Jaramillo C., and Herrera F. (2008) Menispermaceae from the Cerrejón Formation, middle to late Paleocene, Colomiba. Am. J. Bot. 95, 954-973.

Eglinton G and Hamilton RJ, 1967. Leaf epicuticular waxes. Science 156, 1322-1335.

Eglinton G. and Logan G.A. (1991) Molecular preservation. Phil. Trans. Roy. Soc. London B. 333, 315-328.

Freeman K.H. and Colarusso L.A. (2001) Molecular and isotopic records of C4 grassland expansion in the late Miocene. Geochim. Cosmochim. Acta 65, 1439-1454.

Hayes, J.M. (2001) Fractionation of carbon and hydrogen isotopes in biosynthetic processes. Rev. Mineral. Geochem. 43, 225-277.

Herrera F., Jaramillo C., Dilcher D., Wing S.L. and Gomez C. (2008) Fossil Araceae from a Paleocene Neotropical rainforest in Colombia. Am. J. Bot. 95, 1659-1683.

Jaramillo C.A., Pardo-Trujillo A., Rueda M., Harrington G., Bayona G., Torres V. and Mora G. (2007) Palynology of the Upper Paleocene Cerrejón Formation, northern Colombia. Palynology 31, 153-189.

162 Kolattukudy P.E. (1976) The Chemistry and Biochemistry of Natural Waxes. Elsevier, Amsterdam.

Leidelmeyer P. (1966) The Paleocene and Lower Eocene pollen flora of Guyana. Leidse Geologische Mededelingen 38, 49-70.

Lockheart M.J., Van Bergen P.F. and Evershed R.P. (2000) Chemotaxonomic classification of fossil leaves from the Miocene Clarkia lake deposit, Idaho, USA based on n-alkyl lipid distributions and principle component analyses. Org. Geochem. 31, 1223-1246.

Logan G.A., Smiley C.J. and Eglinton G. (1995). Preservation of fossil leaf waxes in association with their source tissues, Clarkia, northern Idaho, USA. Geochim. Cosmochim. Acta. 59(4), 751-763.

Magill C.R., Ashley G.M. and Freeman K.H. (2013). Ecosystem variability and early human habitats in eastern Africa. Proc. Nat. Acad. Sci. 4, 1167-1174.

Osada N., Takeda H., Furukawa A. and Awang M. (2001) Leaf dynamics and maintenance of tree crowns in a Malaysian rain forest stand. J. Ecol. 89, 774-782.

Parker G.G., O’Neill J.P. and Higman D. (1989). Vertical profile and canopy organization in a mixed deciduous forest. Vegatatio 85, 1-11.

Parrish J.T. (1998) Interpreting pre-Quaternary climate from the geologic record. Columbia University Press, New York.

Phillips D.L. and Koch P.L. (2002) Incorporating concentration dependence in stable isotope mixing models. Oecologia 130, 114-125.

Rommerskirchen R., Plader A., Eglinton G., Chikaraishi Y. and Rullkötter J. (2006) Chemotaxonomic significance of distribution and stable carbon isotopic composition of long-chain alkanes and alkan-1-ols in C4 grass waxes. Org. Geochem. 37, 1303-1332.

Schouten S., Woltering M., Rijpstra W.I.C., Slujis A., Brinkhuis H. and Sinninghe Damsté J.S. (2007). The Paleocene-Eocene carbon isotope excursion in higher plant organic matter: Differential fractionation of angiosperms and conifers in the Arctic Earth Planet. Sci. Lett. 258(3-4), 581-592.

Smith F.A., Wing S.L. and Freeman K.H. (2007) Carbon and hydrogen isotope compositions of plant lipids during the PETM as evidence for the response of terrestrial ecosystems to rapid climate change. Earth Planet Sci. Lett. 262, 50-65.

Tipple B.J. and Pagani M. (2010) A 35 Myr terrestrial higher plant n-alkane stable carbon and hydrogen isotope record from the Gulf of Mexico: implications for North

163 American C4 grasslands and hydrologic cycle dynamics. Earth Planet. Sci. Lett. 299, 250–262.

Van Bergen P.F., Collinson M.E. and Briggs D.E.G. (1995) Resistant biomacromolecules in the fossil record. Acta Bot. Neerlandica 44(4), 319-342.

Vogts A., Moossen H., Rommerskirchen F. and Rullkötter J. (2009) Distribution patterns and stable carbon isotopic composition of alkanes and alkan-1-ols from plant waxes of African rain forest and savanna C3 species. Org. Geochem. 40, 1037- 1054.

Wing S.L., Herrera F., Jaramillo C.A., Gómez-Navarro C., Wilf P. and Labandeira C.C. (2009) Late Paleocene fossils from the Cerrejón Formation, Colombia, are the earliest record of Neotropical rainforest. Proc. Natl. Acad. Sci. U.S.A.106, 18627- 18632.

164 Chapter 6. Shade tolerance, evolution and the ecophysiology of alkane production in angiosperms

6.1. Abstract

Odd-numbered, long-chain n-alkanes (C25-C35) are biomarkers commonly found in sedimentary archives and widely used to reconstruct vegetation and climate parameters of the past. Alkanes are derived from leaf cuticle waxes and are produced primarily by angiosperms. The amount of alkanes produced by a leaf is affected by ontogeny and environment, as well as phylogeny. The purpose of these metabolites is not well understood and their proposed function(s) are based on correlation with plant growth conditions. Interpretations of alkanes in the geologic record would benefit from a greater understanding of the taxa, biomes, and environmental conditions associated with alkane production.

We measured the concentration of alkanes and the distribution of alkane chain lengths in a modern closed-canopy forest and found that four of the five sampled genera produced greater amounts of alkanes in understory leaves when compared to mid- and upper- canopy leaves. We also compiled published records of alkane concentrations and distributions in modern leaves and found that core eudictos produce greater amounts of alkanes than basal angiosperms or orders of basal eudicots.

165 The systematic increase in alkane production as well as the higher alkane amounts in understory leaves may be related to the enhanced fungal resistance offered by alkanes that would also be advantageous in the humid understory of a closed-canopy forest.

Increased alkane production in plants of higher angiosperm groups is potentially a component of increased shade tolerance, especially in juvenile trees in closed-canopy forests, and early adaptation by higher angiosperm to closed-canopy conditions.

6.2. Introduction

High-molecular weight, long-chain (C25-C35) n-alkanes are biomarkers commonly used to characterize past terrestrial vegetation and climate (Freeman et al., 1990; Rieley et al.,

1991; Brocks et al., 2003; Pancost & Boot, 2004). Alkanes are replete in the sedimentary record owing to their high preservation potential (Cranwell, 1981). Lacking functional groups, n-alkanes are non-reactive and, particularly those longer than 20 carbons, are rarely subject to microbial degradation, especially in anaerobic conditions (Meyers &

Ishiwatari, 1993; Head et al., 2006; Wentzel et al., 2007).

The odd-numbered carbon chain alkanes n-C25 to n-C35 that are found in modern and ancient sediments are generally attributed to leaf cuticle waxes from vascular plants

(Kolattukudy et al., 1976; Jetter et al., 2000; Kunst & Samuels, 2003), and are used to reconstruct climate (Hinrichs et al., 1998; Schefuss et al., 2003; Pedentchouk et al.,

2008), plant community composition (Gagosian & Peltzer, 1986; Freeman & Colarusso,

2001; Rommerkirschen et al., 2006; Smith et al., 2007; Henderson et al., in review), and

166 even vegetation altitude (Dodd et al., 2003; Polissar et al., 2009). In addition, the carbon isotope compositions are used to distinguish C3 and C4 plant functional type (Rieley et al., 1991; Pagani et al., 2000; Freeman & Colarusso, 2001; Smith et al., 2007; Schouten et al., 2007; Tipple & Pagani, 2010; Magill et al., 2013; Henderson et al., in review). A growing number of studies employ hydrogen isotopic data for n-alkanes to track aridity and hydrologic patterns on ancient landscapes (Liu et al., 2006; Hou et al., 2008; Feakins and Sessions, 2010; Magill et al., 2013).

Angiosperm leaves produce far greater amounts of alkanes than gymnosperms, biasing the record and consequent ecosystem reconstructions toward angiosperm ecophysiology

(Diefendorf et al., 2011). This overproduction is not changed when leaf litter fluxes are considered, as gymnosperm forests produce as much leaf litter as angiosperm forests in similarly structured canopies at the same latitude (Bray & Gorham, 1964; Vogt et al.,

1986; Diefendorf et al., 2011).

Tropical biomes, by geographic expanse and the volume of biomass contributed to sedimentary systems, may dominate the geologic alkane record. Annually, tropical forests produce two to three times the amount of leaf litter that temperate and boreal forests produce, regardless of the taxonomic affiliation of the dominant trees (Bray &

Gorham, 1964; Richards, 1996; Parker et al., 1999; Osada et al., 2001; Graham et al., in review). Litter in tropical regions is also subject to faster decomposition than temperate and boreal forests. These forests dominate the equatorial regions and represent 30% more vegetation coverage, globally, than temperate boreal forests (FAO, 2011). Both

167 temporally and geographically, tropical forests dominate input into sedimentary systems.

Tropical regions cover only 30% of the land mass today but these warmer, wetter biomes were likely much more expansive in the geologic past (Wolfe, 1985; Zeigler et al., 2003).

These observations suggest that a careful examination of alkane production within the context of biome, systematics, and forest structure is necessary to understand the nature of alkane reservoirs, what biomes dominate the alkane record, and possibly expand our understanding of the role of alkanes in leaf physiology.

6.2.1. Ecophysiology and alkane production

n-Alkanes are a major wax component of plant cuticle, the coating found on leaves, stems, flowers, fruits, and seeds. Cuticle consists of a layer of polymeric lipids (such as cutin in aboveground organs and suberin in roots) and layers of soluble wax components

(alkanes, alcohols, ketones, etc). Cuticle thickness varies widely with no apparent phylogenetic trend (Jeffree, 2006). Thickest cuticles are found within orders as divergent as Proteales, Rosales, and Aquifoliales (de Vries, 1968; Sargent, 1976; Heide-Jorgensen,

1978). Cuticle thickness also varies within taxonomic groups. For example, cuticle thickness between members of the genus Hakea can be orders of magnitude different from one another (Heide-Jørgensen, 1980). It is less well understood how alkane amounts differ with higher taxonomic orders.

Cuticle lipids accumulate on the surfaces of leaves over their lifetime, and the relative proportions of lipid compounds change as well (Hauke & Schreiber, 1998; Jetter and

168 Schaffer, 2001). For most plants, cuticle on the adaxial (upper) surfaces of leaves is thicker than on the abaxial (lower) surface (Fisher and Bayer, 1972; Holloway, 1982;

Gordon, 1998). Soluble wax components of the adaxial cuticle are primarily alcohols, and dominant soluble lipids of abaxial cuticle are alkanes (Baker, 1982; Gniwotta et al.,

2005). If cuticle waxes from a portion of a leaf are removed at any point during this accumulation, they are replaced in the same composition as the rest of the leaf (Jetter et al., 2006). This suggests that a leaf’s lipid complement is not a chance reorganization during accumulation but a directed, ontogenetic (growth-related) program of synthesis with certain compounds necessary for certain growth stages (Gao et al., 2012).

Environmental factors such as temperature, humidity, and light can have an effect on the quantity of cuticle waxes and amount and distribution of alkanes in the wax complement of an individual leaf. Experiments with Brassica oleraceae have found that an increase in temperature or decreases in relative humidity can cause an increase in total wax amount with a decrease in the relative concentration of alkanes to total wax (Baker, 1974).

Similarly, Rosa hybrids produce lower amounts of alkanes during drought conditions

(Jenks et al., 2001).

Environmental parameters vary vertically within a forest stand. Therefore, within a single site, trends in cuticular lipid composition reflect three-dimensional forest structure. This study examines the alkane content of leaves sampled from a single site in a tropical rainforest. Leaves were sampled from all heights within the canopy. These leaves

169 represent a wide variety of growth environments along a pronounced light, temperature, and humidity gradient.

We compared our leaf data with published alkane data collected from sun leaves of plants worldwide. This dataset includes the unplaced order Chloranthales, as well angiosperm taxa from the eudicot, monocot, and magnoliid clades. Our goal was to evaluate alkane function and trends in alkane production in the context of both evolution and the development of closed-canopied ranforests.

6.3. Methods

6.3.1. Vertical transect leaf sampling

The Bosque Protector San Lorenzo forest preserve of Panamá (Fig. 1) is located at 9˚

17’N, 79˚38’W. This evergreen, moist tropical terre firme forest is 130 m above sea level and 4.4 km from the Caribbean coast. The site has a tropical monsoonal climate (Köppen classifcation by FAO GeoNetwork) receiving on average 330 cm of rainfall annually and has a mean annual temperature of 26 ºC. The 6 ha preserve contains more than 22,000 trees and 240 recorded species of trees and lianas. The canopy is co-dominated by

Manilkara spp., and Brosimum utile, with many large emergent Calophyllum longifolium throughout the forest. Aspidosperma cruenta and Virola sebifera occupy much of the mid-canopy. (CTFS; http://www.ctfs.si.edu/site/Sherman). This forest exhibits a marked,

170 year-round light gradient with less than 1% of available light (>99% light attenuation) reaching the forest floor. No logging has occurred on this site for at least 200 years and it has been an officially protected site for over a century.

At the San Lorenzo forest, leaves were collected along a 47-m vertical transect using a canopy crane. This tower crane, managed by the Smithsonian Tropical Research Institute

(STRI), provides access to nearly the entire forest structure in a 0.92 ha area. Leaves were selected from all strata of the forest, including the top of the canopy, down to the lowest understory and in gaps. Bulk leaf samples of Aspidosperma cruenta (Apocynaceae),

Brosimum utile (Moraceae), Calophyllum longifolium, (Calophyllaceae), Manilkara spp.(Sapotaceae), and Virola sebifera (Myristicaceae) were taken at 72 unique sampling sites. Leaves were clipped from trees, stored in paper bags, and dried at 70ºC the same day as collection. All samples were acquired during a two-month time window, from

January to February of 2010. After drying, a composite of dried leaves from each sampling site was ground and homogenized. Temperature, humidity, a measurement of the percentage of daily available light captured at a sampling site were collected for all leaf samples (Graham et al., in review).

6.3.2. Alkane extraction and preparative chromatography

Dried ground leaf material (0.5 g) was extracted using a Dionex Accelerated Solvent

Extraction system with dichloromethane/methanol (65:35, v/v). Samples were heated for

5 min at 76 bar and 100ºC, followed by a 5 min heated hold, flushed with solvent mix

171 and purged with N2 for 150 s. This process was repeated three times. Total lipid extract are chromatographically separated into polarity fractions with 0.5 g of high purity silica gel with a 60 Å pore size and 70-230 mesh particle size. n-Alkanes were eluted in 1:20 dichloromethane:hexanes while more polar compounds were eluted with ethyl acetate, dichloromethane and methanol.

6.3.3 Alkane identification and quantification

Extracted alkanes were identified using a Hewlett-Packard (HP) 6890 gas chromatograh

(GC) connected to an HP 5973 quadropole mass spectrometer (MS). We used a fused silica capillary column (Agilent J&W DB-5; 30 m, 0.32 mm, 0.25 µm) with helium carrier gas. Injections were made in hexanes and at an inlet temperature of 320ºC. The column flow rate is 2.0 ml/min and the column oven starts with an initial temperature of

60ºC for 1 min, followed by a ramp to 320ºC at 6ºC/min. This final temperature is then held for 20 min. The MS was operated with a scanning mass range of m/z 50-700 at 3 scans per second and an ionization energy of 70 eV. Compounds are identified using standards, the NIST 98 spectrum library, fragmentation patterns, the published literature, and chromatographic retention times.

Compounds were quantified on a HP 5890 GC with a flame ionization detector (FID) under the same GC conditions as for identification with the same type of column.

Compound peak areas were normalized to co-injected internal standard containing alkane nC-16 and 1,1’-binaphthyl. Aliquots were converted to mass quantities using response

172 curves for 28 n-alkanes ranging from n-C12 to n-C40 analyzed in concentrations ranging from 0.2 to 200 µg/ml. Concentrations were normalized to the mass of dry leaf, fossil or rock material extracted. Average n-alkane chain lengths are calculated using abundances with the following formula modified from Eglinton & Hamilton (1967)

ACL29-35 = ((C29*29) + (C31*31) + (C33 *33) + (C35 *35))/(C29+ C31+ C33+ C35)

where C29 - C35 represents the measured abundance for alkane homologue. This method is modified to reflect the statistically similar production of n-heptacosane (alkane n-C27) common to all taxa (Henderson et al., in review).

6.3.4. Published data selection

We combined data published from 11 regional studies into a large dataset of terrestrial angiosperm leaf n-alkane concentration data (n = 230) to evaluate trends in alkane concentration and distribution patterns by taxa. The dataset includes representatives of

57 families in 22 orders across 11 clades including C3 woody shrubs, trees, and vines (n=

116), C3 grasses and herbs (n=27), and C4 grasses and herbs (n=62). The dataset also includes 15 gymnosperm trees. Samples come from North America (n=44), South

America (n=33), Africa south of the equator (n=73), Africa north of the equator (n=57), and Asia (n=33), Australia (n=1). We use JMP (SAS, 2012) software and R (R

Development Core Team, 2008) to statistically analyze the data. n-Alkane concentration data is not normally distributed so JMP (SAS, 2012) is used to log transform data for

173 analyses. Data for individual alkane homologues is used to calculate ACL29-35 using the method previously outlined and summed for total concentration.

6.4. Results

6.4.1. San Lorenzo, Panama

The biomass-weighted amount of n-alkanes in canopy leaves is highly variable among taxa and through the canopy. Total alkane concentrations for the five genera sampled ranged from 209 µg/g to 2702 µg/g. Overall Aspidosperma had the highest concentration of alkanes, while Calophyllum concentrations had the lowest. When binned by understory versus canopy growth habit a trend emerges in alkane amounts. Understory leaves contain generally greater amounts of alkanes in all but one genus. Figures 2a – 2e show the concentration and distribution for all alkane chains ≥ n-C27 for leaves that receive less than 40% of available daily irradiance (dark bars) compared with those that receive greater than 40% of daily available irradiance. The relative humidity in the understory ranges between 85-97% while the canopy humidity is 65-85%.

Average chain lengths for the five genera range between 29.1 to 32.6, with no clear trend with humidity or light. Binned understory leaves did not have lower ACL in all five taxa despite expectations from work by Giese (1975). This confirms findings by Baker (1974) that light did not affect wax component production. One-way ANOVA found that only

174 binned leaves from Manilkara and Calophyllum had ACL values statistically different from other leaves (F(1,55) = 9.965, p = 0.00227) and (F(1,55) = 7.893, p = 0.00294) respectively. Calophyllum in particular contain little to no alkanes with chain lengths ≥

C29 and Manilkara had little to no C27. Virola leaves contained no ≥ C35. While ACL was not necessarily diagnostic these broad taxonomic patterns were persistent in leaves throughout the canopy.

6.4.2. Published data

Alkane concentrations in a single taxonomic group vary widely. For example, individuals within Asterales expressed alkane concentrations ranging from 62 to 2573 µg/g (Fig. 4).

Leaf samples from the canopy transect also display similar ranges (see Figures 2a – e).

Although ranges overlap, means are distinctive between groups (see Table of p-values).

Magnoliids, Arecales, Proteales, and Cornales contained statistically lower mean concentrations of alkanes when compared with grasses and eudicots (other than

Cornales). Unlike the canopy transect collection, data from the literature show that higher alkane concentrations were not limited to understory plants or low stature trees and samples with low alkane concentrations included sun leaves from herbs, shrubs, and trees.

Similar to the canopy transect collection, published alkane data yielded ACL values for taxonomic groups that were indistinguishable from one another (Fig. 3). All clades overlapped and ranges were often very large. For example the ACL range of asterids was

175 three units and the range of the order Poales alone was two units. This observation confirms that widely divergent species can have very similar alkane chain-length distributions, while plants of the same species or even the same organism can have a very different distribution. This is consistent with work that finds growth-related (ontogenetic) patterns of alkane accumulation (Eglinton & Hamilton, 1963; Baker, 1982; Pancost &

Boot, 2004).

6.5. Discussion

6.5.1. Closed canopies and the early angiosperms

Results from the modern forest study and analysis of the published data find that shaded understory leaves and leaves from the more specialized core eudicot taxa of angiosperms both produce greater amounts of alkanes. This observation begs the question of what is common, ecologically or physiologically, between these groups that they would have similar leaf lipid expression (see Fig. 4).

Multi-stratal forests with closed-canopies and deeply shaded understories have been dominated, and possibly defined by, angiosperms (Upchurch & Wolfe, 1987; Morley,

2000). Prior to the expansion of angiosperms in the Early Tertiary the low-light niches were likely filled by pteridophytes (ferns) and cycads (Crane, 1987; Coiffard et al.,

2007). The habit and extent of Mesozoic pteridophytes is poorly understood, however,

176 and largely based on analogy to extant taxa (Hughes, 1994; Page, 2002;). Most extant pteridophytes are from a lineage that diversified after the rise of angiosperms and adapted to low-light and epiphytic niches created by multi-stratal angiosperm forests (Schneider et al., 2004; Schuettpelz & Pryer, 2009).

Angiosperms may have evolved in, and have become specialized for low-light, humid, environments (Upchurch & Wolfe, 1987; Taylor & Hickey, 1996). Basal angiosperms mastered key innovations such as large seeds, the ability to respond to low and variable light, and protection from a variety of folivores that enabled them to occupy low-light understory niches (Tiffney, 1984; Feild et al., 2003; Royer et al., 2009). More specialized taxonomic orders of angiosperms (such as eudicots) retain these characteristics but have also developed advantageous reproductive and hydraulic features that allow them to dominate many niches, including the understory (Thorne, 1996). The subsequent diversification of angiosperm families has led to the development of the stratified canopy

(Upchurch & Wolfe, 1987).

6.5.2. The function of alkanes on leaf surfaces

The relationship between canopy height, light, and alkane concentrations in our study supports earlier work that found a relationship between humidity and the total amount of alkanes produced in cuticle waxes (Baker, 1974; Jenks & Ashworth, 1999). This increase in alkane concentration in these studies was thought to be a response to drought, yet the

177 same increases in leaf alkane concentration can be found within individual trees from our canopy transect data for sampled leaves with very different humidity.

Greater alkane concentration and longer chain-length alkanes have been found to decrease water vapor permeability in leaves by enhancing cuticle crystallinity (Riederer

& Schreiber, 1995; Kerstiens, 1996; Burghardt & Riederer, 2006; Schreiber, 2006;

Kosma & Jenks, 2007; Leide et al., 2007). The increase in alkanes on the abaxial leaf surface – where most stomata are located – may then be necessary for desiccation prevention. In the leaves from Panamá, long-chain homologues did not increase in the upper canopy, where leaves experience greater water stress, suggesting higher alkane concentrations were unrelated to the influence of cuticle crystallinity (Riederer &

Schreiber, 2006). In the geochemical literature authors have attributed higher alkane amounts and longer alkane chain lengths to greater temperature and/or aridity (Eglinton

& Hamilton, 1962; Pancost & Boot, 2004; Freeman & Pancost, 2013) which may be a reflection of the influences of waxes from grasses (Vogt et al., 1986; Henderson et al.,, in review) The highest concentrations of alkanes in the Panamá dataset, which did not include any grasses, were from leaves in the warmest, most humid strata of the forest.

Examples of leaves that have either been bred to produce greater alkane concentrations on the surfaces of leaves or that have naturally occurring high concentrations of alkanes are found to have increased resistance to fungal infection, especially from rusts and powdery mildews (Gniwotta et al., 2005; Goodwin and Jenks, 2005; Uppalapati et al.,

2012). For example, abaxial surfaces of Hedera leaves have cuticular waxes with an

178 alkane content of 70% of the total wax complement. Fungi on these surfaces have 80% less hyphal germination efficiency than fungi on adaxial surfaces of the leaves which contain 71% alcohols in the total wax complement (Goodwin and Jenks, 2005).

Similarly, Medicago mutants that produce less overall cuticle wax but much greater concentrations of alkanes acquire resistance to many strains of rusts (Uppalapati et al.,

2012). Rose cultivars bred to have increased alkane production in cuticle suffer one-third the fungal infection that wild-type cultivars suffer (Jenks et al., 2001). The abaxial surface is the site of the majority of leaf stomata and the most prone to fungal invasion through these openings. Further, leaf and fruit fusion caused by fungal infection has been found to be directly related to low alkane concentration (Chen et al., 2003; Kurata et al.,

2003).

6.5.3. Alkane concentrations related to other defense compounds in leaves

While rosid and asterid eudicot taxa may produce greater amounts of cuticular alkanes, all higher terrestrial plants, both gymnosperms and angiosperms, have the ability to produce alkanes. The chemical pathway by which alkanes in plants are produced

(decarbonylation of an acyl intermediate) is also common to some algae (Kolattukudy et al., 1973; Wang & Kolattukudy, 1995). Alkanes are not the only specialized (secondary) metabolites of plants that are both constitutive and defensive functions. Other constitutive antifungal compounds common in terrestrial plants include alkaloids, aromatics, terpenoids, and other aliphatics (Grayer & Harbone, 1994).

179 As a class of compounds, terpenoids are common to both gymnosperms and angiosperms but gymnosperms produce almost exclusively diterpenoids while angiosperms produce primarily pentacyclic triterpenoids. Terpenoids can be synthesized either by the mevalonic acid pathway or the 2-C-methy-D-erythritol-4 phosphate pathway and both pathways are found in both broad groups of plants, gymnosperms and angiosperms, as well as the Bacteria (Lange et al., 2000; Tholl and Lee, 2011). While terpenoid amounts can vary in response to pathogen exposure, gymnosperms consistently produce two to three times more terpenoids than angiosperms (Diefendorf, 2010). This observation might suggest that there is a trade-off between compounds that serve the same function. The exceptions to this generalization can be examined for a possible chemotaxonomic or phytogeographic pattern in alkane:terpenoid production.

Our compliation of literature data includes samples from three orders recognized as magnoliids in the current APG III phylogeny (APG, 2009). All three of these orders had very small concentrations of leaf alkanes. Our study also includes a vertical transect of alkane concentrations in Virola, a genus in the family Magnoliales, which had no statistical difference in alkane concentration in understory vs. canopy leaves. Early angiosperms and magnoliids produce benzylisoquinoline alkaloids (Wink & Roberts,

1998). Eight eudicot orders can produce benzylisoquinolines but only one genus

(Erythrina in Fabaceae) accumulate these compounds (Wink et al., 2010). These compounds has been confirmed in all but a one family in the magnoliid clade (Liscombe et al., 2005; Wink, 2010; Hagel & Facchini, 2013). Benzylisoquinolines have been found to have broad antifungal and antimicrobial properties (Wink 1993; Facchini, 2001). In

180 addition to these alkaloids Chloranthaceae produce large amounts of terpenoids and possess significant resistance to pathogens (Kirschner et al., 2010). Many magnoliid leaves have uniquely-shaped spherical glands on both adaxial and abaxial surfaces that secrete complex mixtures of aromatics, resins and mucilage (Walters and Keil, 1996).

A few orders of basal and unranked core eudicots - Proteales, Ranunculales, and

Berberidopsidales -also produce benzylisoquinoline alkaloids (Liscombe et al., 2005).

Considering data for basal and unranked core eudicot groups, including Proteales,

Saxifragales, Caryophyllales, and Santalales, we find Proteales had very small amounts of alkanes. Saxifragales, the most basal rosid, also produced very small amounts of alkanes but the other orders of basal eudicots produced alkanes with mean concentrations in the same range as higher eudicots (see Figure 4) (Burleigh et al., 2009). Likewise,

Cornales, the most basal asterid order, also produces the smallest amount of alkanes

(Bremer, 2009). This observation follows the trend that more basal lineages have leaves with low alkane concentrations and suggests they rely on other defense compounds for fungal protection such as the indoles found in Cornales and the flavinoids and myrecetin found in Saxifragales (Wink, 1993).

Both orders of monocots contain lower concentrations of leaf alkanes when compared to rosids and asterids. Poales have a wide array of constitutive antifungal compounds ranging from saponins to alkaloids, fatty acids, and phenolics. Cultivated genera have been found to produce constitutive antifungals instead of those induced by fungal infection (Crombie et al., 1984; Wippich & Wink, 1985; Jambunathan et al., 1986; Kato

181 et al., 1993). Monocots in general share many of the same defense compounds and synthetic pathways (Wink 1988; Kato et al., 1993). Similarly, Arecales produce saponins and flavonoids (Dahlgren et al., 1985).

While Poales produce, overall, lower amounts of alkanes than eudicots, grasses have greater concentrations of long-chain (n-C33 and n-C35) (Henderson et al., in review). This could be an adaptation to drought stress since longer-chain n-alkanes are associated with lower rates of water loss in leaves (Wilkinson & Mayeux, 1990; Kosma & Jenks, 2007).

Longer chain-length alkanes may decrease permeability by enhancing cuticle crystallinity, creating wax plugs or tortuous paths from stomata to leaf surface, which averts fungal hyphae penetration (Riederer & Schreiber, 1995; Kerstiens, 1996;

Burghardt & Riederer, 2006; Schreiber, 2006; Kosma & Jenks, 2007; Leide et al., 2007).

6.5.4. Phylogenetics and metabolite production

Many secondary metabolites, such as benzylisoquinolones, are expressed only in certain taxa and in some cases such taxa are not phylogenetically related. The “patchy” distribution of metabolite expression in unrelated taxa may be due to convergent evolution or may indicate that the genes responsible for these metabolites are silenced in certain groups. Alkanes are unlike secondary metabolites since alkanes exhibit universal occurrence in terrestrial plants but there is preferential production among certain taxa.

182 At the Panama forest site, we found widely different amounts of alkanes in leaves from the same organism. In particular, leaf-wax concentrations differed greatly between upper canopy and understory leaves, which suggest growth environment can affect alkane production. The canopies of tropical rainforest have strong gradients in temperature, humidity, light and represent diverse habitats for folivores. While any one or a combination of these parameters could be influential in the alkane production patterns experiments in plant breeding suggest humidity is the factor most likely to drive alkane production (Gniwotta et al., 2005; Goodwin and Jenks, 2005; Uppalapati et al., 2012).

Both experiments and phylogenetic comparisons indicate that alkanes can function as defense compounds against fungal infection.

Future work to test the phylogenetic patterns in metabolite expression would benefit from focus on the distribution of constitutive compound production in the basal plant lineages rather than the well-studied higher angiosperm groups. Future studies should also consider the wide array of natural ecosystems that encompass environmental gradients

(such as forest canopies) in order to understand the taxa that specialize in these conditions and the effects of environmental conditions on metabolite production.

6.6 Conclusion

Recent work indicates that n-alkanes can serve as an anti-fungal agent at that increased production of alkanes in waxes proffers a distinct defensive advantage. Observations in natural and laboratory settings have found that leaves in conditions of higher humidity

183 produce higher concentrations of alkanes in epicuticular leaf waxes. Our analysis of leaves from a vertical transect of closed-canopy forest also found a correlation between alkane concentration and humidity. Further, an analysis of the literature also finds that there is a complicated but clear tendency for more specialized orders of angiosperms to also produce greater concentrations of alkanes.

The correlation between antifungal protection, humidity and phylogeny could indicate that alkanes have served as an angiosperm adaptation to shade associated with the rise of closed-canopy forest structure. Further sampling is necessary across a wider range of taxa in order to confirm this observation. While alkanes can enhance tolerance to water stress there is little need for this utility in the humid understory of a closed-canopy rainforest.

The antifungal properties of alkanes promote production in shaded habitats. The systematic increase in alkane production and the increased shade tolerance among angiosperms could be related to the multiple functionality of alkanes.

184 6.7 Figures

Figure 6.1. Closed-canopy field site for vertical sampling of leaf material and light measurement. Digital elevation model by Amante & Eakins (2009).

185

Figure 6.2a. Abundance of extracted alkanes in dry leaf material from Aspidosperma cruenta (Gentianales). Dark bar denotes concentrations for leaves collected in conditions with less than 40% available light (understory). Lighter bar denotes mid- and upper- canopy leaves in ≤ 40% available light.

186 Figure 6.2b. Abundance of extracted alkanes in dry leaf material from Brosimum utile (Rosales). Dark bar denotes concentrations for leaves collected in conditions with less than 40% available light (understory). Lighter bar denotes mid- and upper-canopy leaves in ≤ 40% available light.

187

Figure 6.2c. Abundance of extracted alkanes in dry leaf material from Calophyllum longifolium (Malpighiales). Dark bar denotes concentrations for leaves collected in conditions with less than 40% available light (understory). Lighter bar denotes mid- and upper-canopy leaves in ≤ 40% available light.

188 Figure 6.2d. Abundance of extracted alkanes in dry leaf material from Manilkara spp. (Ericales). Dark bar denotes concentrations for leaves collected in conditions with less than 40% available light (understory). Lighter bar denotes mid- and upper-canopy leaves in ≤ 40% available light.

189 Figure 6.2e. Abundance of extracted alkanes in dry leaf material from Virola sebifera (Magnoliales). Dark bar denotes concentrations for leaves collected in conditions with less than 40% available light (understory). Lighter bar denotes mid- and upper-canopy leaves in ≤ 40% available light.

190 Figure 6.3. Average chain length of extracted alkanes in leaves from a compilation of published data, arranged by broad taxonomic association, ordered from unranked angiosperms to core eudicots and binned by orders or eudicot clades.

191 Figure 6.4. Abundances of extracted alkanes in leaves from a compilation of published data arranged by broad taxonomic association ordered from unranked angiosperms to core eudicots and binned by orders or eudicot clades. Angiosperm phylogeny is constructed following The Angiosperm Phylogeny Group (APG III, 2009).

192 6.8. References

Baker E.A. (1974) The influence of environment on leaf wax development in Brassica oleraceae var. gemmifera. New Phytologist 73, 955-966.

Baker E.A. (1982) Chemistry and morphology of plant epicuticular waxes. In: The Plant Cuticle (eds. D.J. Cutler, K.L. Alvin and C.E. Price) Academic Press, London. pp. 139-165.

Bray J.R. and Gorham E. (1964) Litter production in forests of the world. Advan. Ecol. Res. 2, 101-157.

Bremer B. (2009) Eukaryota; Viridiplantae; Streptophyta; Magnioliophyta; Asterids. In: The Timetree of Life (eds. S.B. Hedges and S. Kumar). Oxford University Press, Oxford, UK. pp. 1177-187.

Brocks J.J., Summons R.E., Heinrich D.H. and Karl K.T. (2003) Sedimentary hydrocarbons, biomarkers for early life. In: Treatise on Geochemistry. Pergamon, Oxford UK. pp. 63–115.

Burghardt M. and Riederer M. (2003) Ecophysiological relevance of cuticular transpiration of deciduous and evergreen plants in relation to stomatal closure and leaf water potential. J. Exp. Bot. 54, 1941-1949.

Burleigh J.G., Hilu K.W., and Soltis D.E. (2009) Inferring phylogenies with incomplete data sets: A 5-gene, 567 taxon analysis of angiosperms. BMC. Evol. Biol. 9, 61- 1949.

Chen X., Goodwin S.M., Boroff V.L., Liu X. and Jenks M.A. (2003) Cloning and characterization of the WAX2 gene of Arabidopsis involved in cuticle membrane and wax production. Plant Cell 15, 1170-1185.

Coiffard C., Gomez B. and Thevenard F. (2007) Early Cretaceous angiosperm invasion of Western Europe and major environmental changes. Ann. Bot. 100(3), 545-553.

Crane P.R. (1987) Vegetational consequences of the angiosperm diversification. In: The Origin of Angiosperms and their Biological Consequences (eds. E.M. Friis, W.G. Chaloner and P.R. Crane) Cambridge University Press, Cambridge. pp. 107-144.

Cranwell P.A. (1981) Diagenesis of free and bound lipids in terrestrial detritus deposited in a lacustrine sediment. Org. Geochem. 3(3), 79-89.

Crombie W.M.L., Crombie L., Green J.B. and Lucas J.A. (1986) Pathogenicity of the take-all fungus to oats: in relationship to the concntration and detoxification of the four avenacins. Phytochem. 27, 3105-3107.

193

Dahlgren R.M.T., Clifford H.T. and Yeo P.F. (1985) The families of the monocotyledons. Structure, Evolution and Taxonomy. Springer-Verlag, New York, NY.

De Vries H.A.M. (1968) Development of the structure of the normal, smooth cuticle of the apple ‘Golden delicious’. Acta Bot. Neerlandica 17, 229-241.

Diefendorf A.F., Freeman K.H., Wing S.L. and Graham H.V. (2011) Production of n- alkyl lipids in living plants and implications for the geologic past. Geochim. Cosmochim. Acta 75, 7472-7485.

Dodd R.S. and Poveda M.M. (2003). Environmental gradients and population divergence contribute to variation in cuticular wax composition in Juniperus communis. Biochem. Systematic Ecol. 13, 1257-1270.

Eglinton G. and Hamilton R.J. (1967) Leaf epicuticular waxes. Science 156, 1322-1335.

Facchini P.J. and Sessions A.L. (2001) Alkaloid biosynthesis in plants: Biochemistry, cell biology, molecular regulation, and metabolic cngineering applications. Ann. Rev. Plant Physiol. 52, 29-66.

Feakins S.J. and Sessions A.L. (2010) Controls on the D/H ratios of plant leaf waxes from an arid ecosystem. Geochim. Cosmochim. Acta 74(7), 2128-2141.

Field T.S., Franks P.J. and Sage T.L. (2003) Ecophysiological shade adaptation in the basal angiosperm, Austrobaileya scandens (Austrobaileyaceae). Int. J. Plant Sci. 164, 313-324.

Fisher D.A. and Bayer D.E. (1972) Thin sections of plant cuticles, demonstrating channels and wax platelets. Canadian J. Bot. 50, 1509-1511.

Food and Agriculture Organization, United Nations (2011) State of the World’s Forests. Accessed on July 2013 from www.fao.org/docrep/013/i2000e.htm

Freeman K.H., Hayes J.M., Trendel J.M. and Albrecht P. (1990) Evidence from carbon isotope measurements for diverse origins of sedimentary hydrocarbons. Nature 343, 254-256.

Freeman K.H. and Colarusso L.A. (2001) Molecular and isotopic records of C4 grassland expansion in the late Miocene. Geochim. Cosmochim. Acta 65, 1439-1454.

Gagosian R.B. and Peltzer E.T. (1986) The importance of atmospheric input of terrestrial organic material to deep-sea sediments. Org. Geochem. 10, 661-669.

194 Gao L., Burnier A. and Huang Y. (2012) Quantifying instantaneous regeneration rates of plant leaf waxes using stable hydrogen isotope labeling. Rapid Commun. Mass Spectrom. 26(1), 115-122.

Gniwotta F., Vogg G., Gartmann V., Carter T.L.W., Riederer M. and Jetter R. (2005) What do microbes encounter at the plant surface? Chemical composition of pea leaf cuticular waxes. Plant Physiol. 139, 519-531.

Goodwin M.A. and Jenks M.A. (2005) Plant cuticle function as a barrier to water loss. In Plant Abiotic Stress (eds. M.A. Jenks and P.M. Hasegawa) Blackwell Publishing, Cambridge, UK. pp. 14-36.

Gordon D.C., Percy K.E. and Riding R.T. (1986) Effect of enhanced UV-B radiation on adaxial leaf surface micromorphology and epicuticular wax biosynthesis of sugar maple. Chemosphere. 36, 853-858.

Grayer R.J. and Harbone J.B. (2013) A survey of antifungal compounds from higher plants. Phytochem. 37, 19-42.

Hagel J.M. and Facchini F.J. (2013) Benzylisoquinoline alkaloid metabolism: A century of discovery and a brave new world. Plant Cell Phyisol. 54, 647-672.

Hauke, V. and Schreiber, L. (1998) Ontogenetic and seasonal development of wax composition and cuticular transpiration of ivy (Hedera helix L.) sun and shade leaves. Planta 207, 67-75.

Head I.M., Jones D.M. and Roling W.F. (2006) Marine microorganisms make a meal of oil. Nat. Rev. Microbiol. 4, 173-182.

Heide-Jørgensen H.S. (1978) The xeromorphic leaves of Hakea suaveolens R. Br. III. Structure of epidermal cells, cuticle developments and ectodesmata. Botanisk Tidsskrift. 72, 227-244.

Heide-Jørgensen H.S. (1980) Cuticle development and ultrastructure – evidence for a procuticle of high osmium affinity. Planta. 183, 511-519.

Hinrichs K.-U., Rinna J., Rullkötter J. (1998) Late Quaternary palaeoenvironmental conditions indicated by marine and terrestrial molecular biomarkers in sediments from the Santa Barbara basin . In: Proceedins of the 14th Annual Pacific Climate Workshop (eds. R.C. Wilson and V.L. Tharp) pp. 125-133.

Holloway, P.J. (1984) Surface lipids of plants and animals In: CRC Handbook of Chromatography: Lipids 1 (eds. H.K. Mangold, G. Zweig and J. Sherma). CRC Press

195 Hughes N.E. (1994) The enigma of angiosperm origins Cambridge University Press, Cambridge.

Hou J., D’Andrea W.J. and Huang Y. (2008) Can sedimentary leaf waxes record D/H ratios of continental precipitation? Field, model, and experimental assessments. Geochim. Cosmochim. Acta 72(14), 3503-3517.

Jambunathan R., Butler L.G. Bandyopadhyay R. and Mughogho L.K. (1986) Polyphenol concentrations in grain, leaf, and callus tissues of mold-susceptible and mold- resistant sorghum cultivars. J. Agric. Food Chem. 34(3), 425-429.

Jeffree C.E. (2006) The fine structure of the plant cuticle . In: Biology of the Plant Cuticle (eds. M. Riederer and C. Müller). Blackwell Publishing, Oxford, UK. pp. 11-125.

Jenks M.A. and Ashworth E.N. (1999) Plant epicuticular waxes: Function, production, and genetics. Hortic. Rev. 23, 1-68.

Jenks M.A., Andersen L., Teusink R.S. and Williams M.H. (2001) Leaf cuticular waxes of potted rose cultivars as affected by plant development, drought and paclobutrazol treatments. Physiolog. Plantarum 112, 62-70.

Jetter, R. and Schäffer, S. (2001) Chemical composition of the Prunus laurocerasus leaf surface. Dynamic changes of the epicuticular wax film during leaf development. Plant Physiol. 126, 1725-1737.

Kato H., Kodama O. and Akatsuka T. (1993) Oryzalexin-E, a diterpene phytoalexin from UV-irradiated rice leaves. Phytochem. 33(1), 79-81.

Kerstiens G. (1996) Cuticular water permeability and it’s physiological significance. J. Exp. Bot. 47, 1813-1832.

Kirchner, K., Wisnlewski A., Cruz A.B., Biavatti M.W. and Netz D.J.A. (2010) Chemical composition and antimicrobial activity of Hedyosmum brasiliense Miq., Chloranthaceae, essential oil. Brazilian J. Pharmacol. 20(5), 692-699.

Kolattukudy P.E. Buckner J.S. and Liu T.-Y. (1973) Biosynthesis of secondary alcohols and ketones from alkanes. Arch. Biochem. Biophys. 156, 613-620.

Kolattukudy P.E. (1976) The Chemistry and Biochemistry of Natural Waxes. Elsevier, Amsterdam.

Kosma D.K. and Jenks M.A. (2007) Eco-physiological and molecular-genetic determinants of plant cuticle function in drought and salt stress tolerance. In: Advances in Molecular Breeding toward Drought and Salt Tolerant Crops (eds.

196 M.A. Jenks P.M. Hasegawa and S.M. Jain). Springer, Dordrecht, The Netherlands pp. 91-120.

Kunst L. and Samuels A.L. (2003) Biosynthesis and secretion of plant cuticular wax. Prog. Lipid Res. 42, 51-80.

Kurata T., Kawabata-Awai C., Sakuradani E., Shimizu S., Okada K. and Wada T. (2003) The YORE-YORE gene regulates multiple aspects of epidermal cell differentiation in Arabidopsis. Plant J. 36, 55-66.

Lange B.M., Rujan T., Martin W. and Croteau R. (2000) Isoprenoid biosynthesis: The evolution of two ancient and distinct pathways across genomes Proc. Nat. Acad. Sci. 97(24), 13172-13177.

Leide J., Hildebrant U., Reussing K., Riederer M. and Vogg G. (2007) The developmental pattern of tomato fruit wax accumulation and its impact on tucicular transpiration barrier properties: Effects of a deficient in a β-ketoacyl- coenzyme A Synthase (LeCER6). Plant Physiol. 144(3), 1667-1679.

Liscombe D.K. Macleod B.P. Loukanina N., Nandi O.L. and Facchini P.J. (2005) Evidence for the monophyletic evolution of benzylisoquinoline alkaloid biosynthesis in angiosperms. Phytochem. 66(11), 1374-1393.

Liu W., Yang H. and Li L. (2006) Hydrogen isotopic compositions of n-alkanes from terrestrial plants correlate with their ecological life forms. Oecologia. 150(2), 330-338.

Magill C.R., Ashley G.M. and Freeman K.H.. (2013). Ecosystem variability and early human habitats in eastern Africa. Proc. Nat. Acad. Sci. 4, 1167-1174.

Meyers P.A. and Ishiwatari R. (199e) Lacustrine organic geochemistry – an overview of indicators of organic matter sources and diagenesis in lake sediments. Org Geochem 20, 867-900.

Morley R.J. (2000) Origin and Evolution of Tropical Rainforests. John Wiley & Sons, Chichester UK.

Osada N., Takeda H., Furukawa A. and Awang M. (2001) Leaf dynamics and maintenance of tree crowns in a Malaysian rainforest stand. J. Ecol. 89, 774-782.

Pagani M., Freeman K.H. and Arthur M.A. (2000) Isotope analyses of molecular and total organic carbon from Miocene sediments Geochim Cosmochim Acta 64, 37- 49.

Page C.N. (2002) Ecological strategies in fern evolution: a neopteridological overview. Rev. Palaeobot. Palynol. 119, 1-33.

197

Pancost R.D. and Boot C.S. (2004) The palaeoclimatic utility of terrestrial biomarkers in marine sediments. Marine Chem. 92, 239-261.

Parker G.G., O’Neill J.P. and Higman D. (1999). Vertical profile and canopy organization in a mixed deciduous forest. Vegatatio 85, 1-11.

Pedentchouk N., Sumner W., Tipple B. and Pagani M. (2008) δ13C and δD compositions of n-alkanes from modern angiosperms and conifers: an experimental set up in central Washington State, USA. Org. Geochem. 39, 1066-1071.

Polissar P., Freeman K.H., Rowley D., McInerney F.A. and Currie B., (2009) Paleoaltimetry of the Tibetean Plateau from D/H ratios of lipid biomarkers. Earth Planet. Sci. Lett. 287, 64-76.

Riederer M. and Schreiber L. (1995) Waxes: the transport barriers of plant cuticles. In: Waxes: chemistry, molecular biology and functions (ed. R.J. Hamilton). Oily Press, Dundee, UK. pp. 131-156.

Riederer M. and Schreiber L. (2001) Protecting against water loss: Analysis of the barrier properties of plant cuticles. J. Exp. Bot. 52, 2023-2032.

Rieley G., Collier R.J., Jones D.M., Eglinton G., Eakin P.A. and Fallick A.E. (1991) Sources of sedimentary lipids deduced from stable carbon isotope analyses of individual n-alkanes. Nature 352, 425-427.

Richards P.W. (1996) The Tropical Rainforest: An Ecological Study. (2nd ed.) Cambridge University Press, Cambridge UK.

Rommerskirchen R., Plader A., Eglinton G., Chikaraishi Y. and Rullkötter J. (2006) Chemotaxonomic significance of distribution and stable carbon isotopic composition of long-chain alkanes and alkan-1-ols in C4 grass waxes. Org. Geochem. 37, 1303-1332.

Royer D.L., Kooyman R.M., Little S.A. and Wilf P. (2000) Ecology of leaf teeth: A multi-site analysis from an Australian subtropical rainforest. Am. J. Bot. 96, 738- 750.

Sargent C. (1976) In situ assembly of cuticular wax. Planta 129, 123-126.

Schefuss E., Ratmeyer V., Stuut J.B.W., Jansen J.H.F. and Sinninghe Damsté J.S. (2003) Carbon isotope analyses of n-alkanes in dust from the lower atmosphere over the central eastern Atlantic. Geochim. Cosmochim. Acta 67, 1757-1767.

Schneider H., Schuettpelz E., Pryer K.M., Cranfill R., Magallon S. and Lupia R. (2004). Ferns diversified in the shadow of angiosperms. Nature 428, 553-557.

198 Schouten S., Woltering M., Rijpstra W.I.C., Slujis A., Brinkhuis H. and Sinninghe Damsté J.S. (2007). The Paleocene-Eocene carbon isotope excursion in higher plant organic matter: Differential fractionation of angiosperms and conifers in the Arctic Earth Planet. Sci. Lett. 258(3-4), 581-592.

Schreiber L. (2006) Characterisation of polar paths of transport in plant cuticles . In: Biology of the Plant Cuticle (eds. M. Riederer and C. Müller) pp. 280-311.

Schuettpelz E. and Pryer K.M., (2009) Evidence for a Cenozoic radiation of ferns in an angiosperm-dominated canopy. Proc. Nat. Acad. Sci. 106, 11200-11205.

Smith F.A., Wing S.L. and Freeman K.H. (2007) Carbon and hydrogen isotope compositions of plant lipids during the PETM as evidence for the response of terrestrial ecosystems to rapid climate change. Earth Planet Sci. Lett. 262, 50-65.

Taylor D.W. and Hickey L.J. (1996) Evidence for and implications of herbaceous origin for Angiosperms In: Flowering plant origin, evolution, and phylogeny (eds. Taylor D.W. and Hickey L.J.). Chapman and Hall, New York, NY. pp. 232-266.

The Angiosperm Phylogeny Group (2009) An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG III. Bot. J. Linnean Soc. 161, 105-121.

Tholl D. and Lee S. (2011) Terpene specialized metabolism in Arabidopsis thaliana. The Arabidopsis Book 9, e0143.

Thorne R.F. (1996) The least specialized angiosperms In: Flowering plant origin, evolution, and phylogeny (eds. Taylor D.W. and Hickey L.J.). Chapman and Hall, New York, NY. pp. 286-313.

Tiffney B.H. (1984) Seed size dispersal syndromes, and the rise of the angiosperms: Evidence and hypothesis. Ann. MO Bot. Garden 71, 551-576.

Tipple B.J. and Pagani M. (2010) A 35 Myr terrestrial higher plant n-alkane stable carbon and hydrogen isotope record from the Gulf of Mexico: implications for North American C4 grasslands and hydrologic cycle dynamics. Earth Planet Sci. Lett. 299, 250–262.

Upchurch G.R. and Wolfe J.A. (1987) Mid-Cretaceous to early Eocene vegetation and climate: evidence from fossil leaves and woods. In The Origins of Angiosperms and their Biological Consequences (eds. E.M. Friis, W.G. Chaloner and P.R. Crane). Cambridge University Press, Cambridge UK. pp. 75-105.

Uppalapati S.R., Ishiga Y., Doraiswamy V., Bedair M., Mittal S., Chen J., Nakashima J., Tang Y., Tadege M., Ratet PP., Chen R., Schultheiss H. Mysore K.S. (2012) Loss of abaxial leaf epicuticular wax in Medicago truncatula irg1/palm1 mutants

199 results in reduced spore differntiation of anthracnose and nonhost rust pathogens. Plant Cell 24(1), 353-370.

Vogt K.A., Grier C.C and Vogt D.J. (1986) Production, turnover, and nutrient dynamics of above- and belowground detritus of world forests. Advan. Ecol. Res. 15, 303- 377.

Walters D.R. and Keil D.J. (1996) Vascular Plant Taxonomy. Kendall/Hunt Publishing Co., Dubuque, IA.

Wang X. and Kolattukudy P.E. (1995) Solubilization and purification of aldehyde- generating fatty acyl-CoA reductase from green alga Boryococcus braunii. FEBS Lett.370, 15-18.

Wentzel A., Ellingsen T.E., Kotlar. H.K., Zotchev S.B. and Throne-Holst M. (2007) Bacterial metabolism of long-chain n-alkanes. Appl. Microbiol. Biotechnol. 76, 1209-1221.

Wilkinson R. and Mayeux H.S. Jr. (1990) Composition of epicuticular wax on Opuntia engelmannii. Bot. Gazette 151, 342-347.

Wink M. (1988) Plant breeding: importance of plant secondary metabolites for protection against pathogens and herbivores. Theor. Appl. Genet. 75, 225-233.

Wink M. (1993) Allelochemical properties and the raison d’etre of alkaloids. In: The Alkaloids (ed. G. Cordell). Academic Press, London, UK. pp. 1-118.

Wink M. and Roberts M.F. (1998) Molecular modes of action of defensive secondary metabolites. Ann. Plant Rev. 39, 21-161.

Wink M. (2010) Compartmentation of alkaloid biosynthesis, transport and storage. In: Alkaloids: biochemistry, ecology and medicinal applications (eds. M.F. Roberts and M. Wink). Plenum Press, New York, NY. pp. 239-262.

Wippich C. and Wink M. (1985) Biological properties of alkaloids. Influence of quinolizidine alkaloids and gramine on the germination and development of powdery mildew, Ersiphe graminis f. sp. hordei.. Experientia. 41, 1477-1479.

Wolfe J.A. (1985) Distribution of major vegetational types during the Tertiary In: The Carbon cycle and atmospheric CO2 – natural variations, Archean to present (Geophysical Monograph 32) (eds. E.T. Sundquist and W.S. Breker). American Geophysical Union Press, Washington, DC. pp. 357-375.

Ziegler A.M., Eshel G., Rees P.M., Rothfus T.A., Rowley D.B. and Sunderlin D. (2003) Tracing the tropics across land and sea: Permian to present. Lethaia. 36, 227-254.

200

Chapter 7. Research summary

7.1. Chapter summaries

The carbon isotope composition of leaf tissue has been well-studied in relation to environmental conditions and phylogenetic patterns. These efforts have examined leaf material and epicuticular waxes of multiple leaves from the same tree as well as leaves from individuals of different species and genera found in the same biome or geographic area. Very few studies have attempted to characterize the carbon isotope composition and alkane expression in leaves through the array of environmental gradients that naturally occur with height within a canopy. This dissertation seeks to fill that information gap and develop applicable proxy techniques that can help workers query geologic archives to better understand canopy structure in the past. An understanding of how the vertical dimension affects the isotopic and molecular characteristics of leaves also informs our interpretation of soil carbon that arises from leaf litter.

13 Chapter 2 incorporates natural vertical gradients and δ Cleaf values from the San Lorenzo forest to answer how many fossil leaves would be necessary to find the wide range of

13 δ Cleaf values diagnostic of closed-canopy conditions in a paleoflora. This work incorporates biogeochemical data and computational modeling to find the necessary sampling size. The model also predicted the isotopic composition of leaf litter arising from different levels of canopy coverage. Given the greater annual flux of upper canopy

201 leaves into litter soil carbon will reflect the isotopic composition of upper canopy leaves in particular. Our model can be used to demonstrate this flux-dependence in soil carbon systems. Flux and the canopy origin of leaves is an underappreciated dimension of carbon input in soil archives.

13 Chapter 3 is an application of the leaf δ Cleaf range technique outlined in chapter 2 using

Cretaceous and Paleocene fossils from Colombia. Pollen data and floristics at the older flora indicate an open canopy structure while the two Paleocene fossil sites are believed to represent the first closed-canopy Neotropical forest. Our investigation confirms the open-canopy characteristics of the Cretaceous site and also find that the older Paleocene

13 site likely had a closed-canopy. The δ Cleaf range of the younger site was ambiguous but the sedimentary evidence indicates a lacustrine depositional setting. These settings are less likely to express the isotopic range of a closed-canopy forest since most leaves arise from the lakeside and do not experience the environmental and light gradient typical of a closed-canopy, even if the lake is within a closed-canopy forest.

Chapter 4 investigates alkanes preserved in the fossil leaves from the Cretaceous and

Paleocene sites and compares them to leaves of modern relatives. The lipids in sediments adjacent to the fossil were also measured and compared to fossils in order to understand the provenance of alkanes in sediments. While the isotopic ranges of fossil leaves are not the same as modern leaves ∆leaf and εlipid values are comparable to modern leaves indicating similar physiology and carbon metabolism. This investigation finds that

202 depositional conditions and depositional setting greatly affects preservation of alkanes in fossil leaves regardless of taxa.

13 Chapter 5 uses the alkane concentrations and δ Clipid values for fossil leaves found in chapter 4 and combines it with census data from the assemblages in a mixing model that

13 predicts the ACL and δ Clipid values for alkanes isolated from outcrop samples. The model results closely match the lipid profile of bulk rock. Fossil leaves with greater

13 amounts of alkanes were also found to have more depleted δ Cleaf values, consistent with our findings in the San Lorenzo forest which display this same relationship throughout the canopy.

Chapter 6 compares the alkane concentrations in leaves collected throughout the closed- canopy San Lorenzo forest. Leaves in the humid understory contain higher concentrations of alkanes. San Lorenzo data were then compared with a collection of published alkane abundance data that showed a trend towards higher angiosperms producing greater amounts of alkanes. Alkanes exhibit antifungal properties and leaves in humid settings will produce more alkanes, possibly as defense compounds. Alkanes may be produced in greater amounts by higher angiosperms as a defense compound in the humid understory among those taxa that have adapted enhanced shade tolerance.

203 7.2. Future directions

Our research illustrates how ecosystem structure influences the carbon isotope composition of organic carbon in geologic archives is considered. Carbon isotope excursions attributed to ecosystem shifts could indicate a change in canopy structure. Our study also gives new methods for investigating the canopy structure in fossil assemblages. Our methods could be applied to fossils from the PETM of the Big Horn

Basin where canopy structure is believed to be affected by warming (Secord et al., 2008).

13 In fact, this study would confirm the open-canopy nature of the δ Cleaf range of the

Castle Rock flora (Fricke et al., 2007).

Our study indicating that fossil leaves without the pristine preservation of the Clarkia

13 Formation can preserve the alkane concentration and δ Clipid values typical of their modern relatives and their ecosystem niche. While these features may not be unique enough to be chemotaxonomic, these substrates can be queried for physiological evidence of carbon metabolism through ∆leaf and εlipid values. In particular, estimations of εlipid values through time and in ancient taxa can help us understand the controls on carbon isotope fractionation during lipid synthesis.

This study has added greatly to the body of evidence that informs our understanding of the phylogenetic nature of alkane production in angiosperms but much more work is needed. The evidence is interesting, but admittedly thin and only because there is so little data on alkane traits in early angiosperms. We would hope that our results entice others in

204 the community to pursue data collection along these lines and we reiterate pleas from other researchers (Grayer & Harborne, 1994; Wink, 2010) to expand investigations of alkanes to include more ancient lineages and taxa that, while not ecologically important now, enjoyed greater dominance and wider ranges in the geologic past.

205 7.3. References

Fricke H., Ellis B., Johnson K. and Jahren A.H. (2007). Carbon isotope characterization of the closed canopy (?) Castle Rock rainforest. Geological Society of America Abstracts with Programs. 39(6), 27.

Grayer R.J. and Harbone J.B. (2013) A survey of antifungal compounds from higher plants. Phytochem. 37, 19-42.

Secord R., Wing S.L. and Chew A. (2008) Stable isotopes in early Eocene mammals as indicators of forest canopy structure and resource partitioning. Paleobiol. 34,282- 300.

Wink M. (2010) Compartmentation of alkaloid biosynthesis, transport and storage. In: Alkaloids: biochemistry, ecology and medicinal applications (eds. M.F. Roberts and M. Wink). Plenum Press, New York, NY. pp. 239-262.

206 Appendix A. Evapotranspiration, humidity and lipid synthesis

A.1. Introduction

Soil water absorbed by the root does not undergo significant fractionation; isotopic enrichment occurs along the stem between petioles, between leaf and petiole, and from petiole to the leaf tip (Flanagan & Ehleringer, 1991; Luo & Sternberg, 1992). In general, leaf water is greatly enriched in both deuterium and 18O compared to the rest of the plant.

This enrichment arises during transpiration and reflects the lower vapor pressure and binary diffusivity in air of heavier water isotopologues – essentially, ‘heavier’ water is left behind in the leaf during evaporation (Farquhar et al., 2007). Recent work suggests that stomatal conductance and transpiration is driven by radiation (Pieruschka et al.,

2010); other factors affecting leaf-water composition include tissue type, proximity of veination, and stomatal conditions (Yakir & DeNiro, 1989; Flanagan et al., 1991; Luo &

Sternberg, 1992). These anatomic features produce hydraulic and isotopic characteristics that vary both by habitat (Sack et al., 2005) and phylogeny (Helliker and Ehleringer,

2002).

Most important to this study is the greater leaf-water isotopic enrichment observed in sun leaves vs. shade leaves (Yakir et al., 1990; Walker & Lance, 1991). This enrichment translates into the isotopic composition of dry bulk matter and cuticle lipids (Leaney et al., 1985, Chikaraishi et al., 2004; Bi et al., 2005). Carbon-bound hydrogen in straight- chain alkanes does not exchange appreciably over geologic timescales (Sessions et al.,

207 2004; Pedentchouk et al., 2006). Therefore, because of the gradient in transpiration with light exposure, then n-alkane deuterium signatures have an unexplored potential to serve as proxies for canopy coverage, provided synthetic fractionation can be predicted and leaf-water is the hydrogen source (Sternberg, 1988).

We hypothesize that humidity gradients, through their influence on the isotopic composition of leaf waters, are recorded by the δ18O of isolated cellulose. We propose to use this framework for testing how environmental gradients in the canopy influence foliar n-alkane δD compositions.

Transpiration and humidity have been linked to irradiance, temperature and radiation load; these factors are all linked to the light regime and change dramatically through the canopy. We isolated α-cellulose and measured the δ18O values from leaves at multiple sites in order to determine relationships between leaf water and environmental gradients in humidity and transpiration. We also analyzde δD values for extracted n-alkanes from the entire collection and compare results with cellulose and environmental data. These measurements are augmented by environmental data (temperature, percent humidity, etc.) and by deuterium and oxygen isotopic analyses of water vapor collected in air samples.

The cellulose oxygen and lipid deuterium data uniquely allow us to address potential source(s) of hydrogen recorded in lipids (i.e., transpiration-influenced leaf water or a distinct and potentially unfractionated pool; McInerney et al., 2011) within the context of canopy conditions and irradiance regime.

208 A.2. Methods

Leaves from five genera of trees were collected vertically through a forest canopy in

Panama. The daily light environment at each leaf collection site was characterized and humidity and temperature measurements were also recorded (see Chapter2 2). These leaves represent a continuum of evapotranspirative environments within the same organism. Leaves were dried at 70° then ground and homogenized before extraction

A.2.1. n-Alkane isolation

Dried ground leaf material (0.5 g) was extracted using a Dionex Accelerated Solvent

Extraction system with dichloromethane/methanol (65:35, v/v). Samples were heated for

5 min at 76 bar and 100ºC, followed by a 5 min heated hold, flushed with solvent mix and purged with N2 for 150 s. This process was repeated three times. Total lipid extracts were chromatographically separated into polarity fractions with 0.5 g of high purity silica gel with a 60 Å pore size and 70-230 mesh particle size. Hydrocarbons were separated with 100% hexanes, while more polar compounds eluted with ethyl acetate, dichloromethane, and methanol then archived.

A.2.2. Compound specific D/H ratio analysis

Hydrogen isotope ratios of extracted alkanes were measured using an HP 6890 GC connected to a Thermo Finnegan Delta Plus XP IRMS with a combustion interface and a

209 high-temperature pyrolysis furnace held at 1400ºC (Burgoyne & Hayes, 1998). The daily average H3 factor was recorded throughout the analytical time frame. Isotopic composition is determined relative to a reference gas calibrated to an international standard, as well as by regular injections of Mix A (n-C16 to n-C30; Arndt

Schimmelmann, Indiana University) which has been calibrated for D/H ratios. Isotope values of samples are reported in delta notation relative to Vienna Standard Mean Ocean

Water (VSMOW).

A.2.3. α-Cellulose Isolation

The solid sample remainder from lipid extraction was Soxhlet extracted in a 2:1 toluene:ethanol azeotrope, boiled in distilled salt water (1% NaCL), treated with an acetic acid:nitric acid (12:1) mixture, bleached in sodium chlorite and acetic acid (2:1), and finally reacted with sequential solutions of 10% and 17% sodium hydroxide. These steps removed resins, pigments and tannins, water-soluble polysaccharides, starches, and inorganic salts, lignin, and hemicelluloses. Isolated α-cellulose was carefully rinsed in water, ethanol, and acetone between reaction steps and before isotopic analysis. This protocol is an amalgamation of published methods (Leavitt & Danzer, 1993; Roden &

Ehleringer, 1999; Brendel et al., 2000). Care was taken in processing given the tough, highly lipidic nature of our samples.

A.2.4. α-Cellulose δ18O analysis

210 δ18O of solid α-cellulose was measured on a Finnigan MAT 252 by pyrolysis at 1400ºC in a quartz combustion furnace packed with glassy carbon and alumina. CO was separated from other gases chromatographically by a 90ºC molecular sieve column before introduction into the mass spectrometer. Measured δ18O values were corrected for sample size dependency and then normalized for volume to the VSMOW scale with a two-point calibration and internal standards (calibrated with USGS 35 sodium nitrate, 56.81‰;

IAEA NO-3 potassium nitrate, 23.14‰).

A.3. Results

While α-cellulose is not as well preserved on the same geologic timescales as n-alkanes, is useful for estimating the isotopic composition of oxygen in leaf waters based on Craig-

Gordon type models of evaporation (McInerney et al., 2011). We predict a range of

~44‰ in deuterium and 6.5‰ in leaf water based on the measured humidity range (65 to

95%) and ancillary data (T ~28C, the average precipitation composition -34‰, -5.3‰ for

δD and δ18O of water). Our observed δ18O values are consistent with this range, scaled by factors related to cellulose synthesis and by the proportion of unmixed vein water

18 (Pex·Px; Barbour and Farquhar, 2000). Our preliminary δ O values measured for isolated

α-cellulose varied over 6‰ and track observed relative humidity (Figure A1), although notably with offsets for different plant types. In contrast, extracted leaf n-alkanes range about 20‰, but show no pattern in values from canopy top to forest floor (Figure A2).

211 A.4. Discussion

This decoupling between cellulose oxygen and lipid deuterium isotope data for tree leaf samples was unexpected. However, we note grasses grown hydroponically in a greenhouse with soil evaporation prohibited had similar patterns. Cellulose oxygen isotopic signatures follow humidity as expected, but lipid deuterium values did not, and for grasses, lipid and cellulose synthesis potentially tap substrates from isotopically distinct pools (see McInerney et al., 2011). Our preliminary findings suggest that lipid δD values are not sensitive to canopy position. These results suggest that lipid δD data reflect water source composition, which has significant paleoclimate reconstruction utility.

212 A.5. Figures

Figure A1: D/H values for alkane n-C31 expressed as a function of light availability.

Colors represent different genera of trees. While there may be trends within taxonomic groups overall results are insensitive to irradiance.

213

Figure A2: δ18O values for α-cellulose varies with observed relative humidity at each sampling site, with higher humidity sties having more depleted 18O content. Trends are strongly associated with taxonomic groups.

214 A.6. References

Barbour, M.M. and Farquhar, G.D. (2003) Do pathways of water movement and leaf 18 anatomical dimensions allow development of gradients in H2 O between veins and the sites of evaporation within leaves? Plant, Cell and Env. 27, 107-121.

Bi, X., Sheng, G., Liu, X., Li, C. and Fu, J. (2005) Molecular and carbon and hydrogen isotopic composition of n-alkanes in plant leaf waxes. Org. Geochem. 36, 1405- 1417.

Brendel O., Iannetta P.P.M. and Stewart D. (2000) A rapid and simple method to isolate pure α-cellulose. Phytochem. Anal. 11, 7–10.

Burgoyne T.W. and Hayes J.M. (1998) Quantitative production of H2 by pyrolysis of gas chromatographic effluents. Anal. Chem. 70, p. 5136–5141.

Chikaraishi, Y., Naraoka, H. and Poulson, S.R. (2004) Hydrogen and carbon isotopic fractionations of lipid biosynthesis among terrestrial (C3, C4 and CAM) and aquatic plants. Phytochem. 65, 1369-1381

Farquhar, G.D., Cernusak, L.A. and Barnes, B. (2007) Heavy water fractionation during transpiration. Plant Physiol 143, 11-18.

Flanagan, L.B. and Ehlerginer, J.R. (1991) Stable isotope composition of stem and leaf water: applications to the study of plant water use. Functional Ecol. 5, 270-277.

Helliker B.R. and Ehleringer J.R. (2002) Differential 18O enrichment in leaf cellulose of C3 versus C4 grasses. Functional Plant Biol. 29, 435-442.

Leaney F.W., Osmond C.B., Allison G.B., and Zeigler H. (1985) Hydrogen-isotope composition of leaf water in C3 and C4 plants: its relationship to the hydrogen- isotope composition of dry matter. Planta 164, 215-220.

Leavitt S.W. and Danzer S.R. (1993) A method for batch processing small wood samples to holocellulose for stable carbon isotope analysis. Ana.l Chem. 65, 87-89.

Luo, Y. and Sternberg, L. (1992) Spatial D/H heterogeneity of leaf water. Plant Physiol. 99, 348-350.

McInerney F.A., Helliker B. and Freeman K.H. (2011) Hydrogen isotope ratios of leaf wax n-alkanes in grasses are insensitive to transpiration. Geochim. Cosmochim. Acta 75, 541-554.

215 Pedentchouk N., Freeman K.H. and Harris N.B. (2006) Different response of δD values of n-alkanes, isoprenoids and kerogen during thermal maturation. Geochim. Cosmochim. Acta 70, 2063-2072

Pieruschka R., Huber G. and Berry J.A. (2010) Control of transpiration by radiation. Proc. Natl. Acad. Sci. 107, 13372-13377.

Roden J.S. and Ehleringer J.S. (1999) Hydrogen and oxygen isotope ratios of tree ring cellulose for field-grown riparian trees. Oecologia 123, 481–489.

Sack, L. and Holbrook, N.M. (2006) Leaf hydraulics. Ann. Rev. Plant Biol. 57, 361-381.

Sessions A.L., Sylva S.P., Summons R.E. and Hayes J.M. (2004) Isotopic exchange of carbon-bound hydrogen over geologic timescales. Geochim. Cosmochim. Acta 68, 1545-1559.

Sternberg L. (1988) D/H ratios of environmental water recorded by D/H ratios of plant lipids. Nature 333, 59-61.

Walker C.D. and Lance R.C.M. (1991) The Fractionation of 2H and 18O in leaf water of barley. Australian J. Plant Physiol. 18, 411-425.

Yakir, D., and DeNiro, M.J. (1990) Oxygen and hydrogen isotope fractionation during cellulose metabolism in Lemna gibba L. Plant Physiol. 93, 325-332.

216 Appendix B. Nitrogen economics and δ15N response to irradiance

B.1. Introduction

Nitrogen as a proxy of irradiance is primarily due to changes in nitrogen investment between sun vs. shade leaves. Foliar nitrogen content is related to the photosynthetic capacity of the leaf since the majority of leaf nitrogen is incorporated in the proteins and pigments associated with the Calvin cycle and thylakoid membrane compounds.

Approximately 70-80% of leaf nitrogen is held in thylakoid membrane components with

60-80% of this thylakoid nitrogen incorporated into pigments. The remaining foliar nitrogen is incorporated primarily in the ribulose-1,5-bisphosphate enzyme (Evans,

1989). Shaded leaves have lower concentrations of nitrogen than sun leaves but more chlorophyll pigment per unit nitrogen and fewer cytochrome b/f complexes (Bjorkman et al., 1972).

In extant plants, these differences in nitrogen investment can be reflected in the nitrogen content and ratio of C:N of bulk leaf material (Evans and Terashima, 1987;

Evans & Poorter, 2001; Poorter et al., 2006, Valladares & Niinemets, 2008). Fossils do not retain the necessary proportions of organic material to make C:N measurement possible. However, there is an observed link between total % foliar nitrogen content and

15 15 δ N composition of leaf material (Ometto et al., 2006). Further, δ Nleaf values have been found to vary in response to whole plant morphology and canopy position (Martinelli et al., 1999; Ometto et al., 2006; Domingues et al., 2007). Nitrogen isotope expression in

217 15 closed-canopy tropical forests is vastly different than temperate forests. δ Nleaf values could potentially inform interpretation of forest structure and biome in fossil archives.

B.2. Methods

B.2.1. Sample selection and preparation

Bulk leaf samples from five genera of trees were collected at 72 unique sampling sites accessed using the STRI canopy crane or from the forest floor. The daily average of percent available light was characterized at each sampling site. These genera represent the majority of canopy coverage and the greatest inputs into litter assemblages in these forests. Tree selection also represents a wide phylogenetic span within the angiosperms, including magnoliids, rosids, and asterids. Fresh leaves were clipped from trees, stored in paper bags and dried at 70ºC at the Summit Herbarium at STRI. A composite of leaves from each sampling site were ground and homogenized.

B.2.2. Isotopic Analysis

15 Weight percent total organic nitrogen and δ Nleaf was determined using a continuous flow Costech elemental analyzer (EA) connected to a Thermo Finnigan Delta XP IRMS.

Measured δ15N values are routinely corrected for sample size dependence and normalized

218 to the standard atmosphere scale with a two-point calibration (Coplen et al., 2006) using internal standards (calibrated with IAEA N-1 ammonium sulfate, -0.4‰; IAEA N-1 ammonium sulfate, 20.3‰).

B.3. Results

Leaves analyzed in this study show no obvious vertical trend with % foliar nitrogen

(Figure A4) although strong associations with irradiance regime, canopy position, and even generic identity are evident in the nitrogen isotopic composition of sampled leaf material (Figures A5 & A6). Interestingly, upper-canopy leaves exhibit negative δ15N values, consistent with higher pigment turnover rates and greater nitrogen demands than leaves in the understory. This result is quite unexpected since tropical tree leaves generally express positive foliar δ15N whereas temperate tree leaves express negative foliar δ15N.

B.4. Discussion

The different isotopic composition in tropical versus temperate leaves is often interpreted as an indication that tropical forests have a relatively more open nitrogen cycle (higher losses and inputs relative to internal cycling) with greater amounts of nitrogen circulating through these systems annually (Martinelli et al., 1999). This would imply that tropical soils are not nitrogen-limited (Vitousek, 1984). However, our results would suggest that

219 foliar isotopic composition is a response to light environment, a regulation of pigment turnover in order to optimize photosynthetic capacity - rather than a source effect or artifact of soil cycling.

Many authors have assumed that soil nitrogen composition drives foliar δ15N, however there are a number of caveats to consider. Two pathways exhibit fractionating effects: (i) the nitrate reductase pathway responsible for nitrate assimilation and reduction and (ii) the glutamine-to-glutamate synthetase pathway responsible for assimilation of ammonium. Fractionation is only observed when there is a pool of enriched nitrogen leaked from plant roots after uptake during times when demand is high relative to supply

(Evans et al., 1996; Hogberg et al., 1999). Soil nitrogen is dominated by highly recalcitrant pools. Labile nitrogen is transient and often varies widely in isotopic composition over short times and distance from sources (Bergerson et al., 1990). Further, plant uptake is plastic and different taxa tap different pools of nitrogen in different environmental conditions (Hobbie et al., 1998). If the height can be correlated with δ15N this would imply rooting depth has an effect on foliar nitrogen composition. However, tropical tree roots are generally shallow, extending laterally for great distances (Medina

& Minchin, 1980). Association with mycorrhizal fungi can also be a factor except that nitrogen-rich ecosystems such as this forest exhibit less reliance on fungal root infection for nutrient access (Hobbie et al., 2000). Thus, proximity to trees harboring nitrogen- fixing bacteria (such as the Frankia associated with Brosimum) likely have more impact on foliar nitrogen concentration than it’s isotopic composition (Marshall et al, 2007).

220 B.5. Future work and implications

Compound specific analysis of chlorophyll nitrogen could confirm if the unique isotopic signal found in this forest is driven by pigment nitrogen composition. Given the high rates of pigment turnover in the full-sun leaves, we hypothesize positive correlation irradiance, nitrogen demand, and fractionating processes. An association between chlorophyll and light would be useful for interpreting the nitrogen isotopes of geoporphyrins, the taphonomic derivatives of tetrapyrole pigments that can be extracted from fossils and sediments (Dilcher et al., 1970; Niklas & Giannasi, 1978).

221

B.6. Figures

Figure A.4. % Total nitrogen in leaf material does not exhibit any vertical trend with light, height, or taxa.

222

Figure A.4. Nitrogen isotope composition in leaf material trends vertically with height. Upper canopy leaves exhibit a negative δ15N values and understory and mid-canopy leaves exhibit positve δ15N values.

223

Figure A.5. δ15N values arranged by genera. Letters identify points in nitrogen isotpe space and signify the canopy position of the sampled leaves: F = Forest Floor, U = Understory, C = Canopy, and E = Emergent. In general, canopy position increases in height as the δ15N values become more negative. This plot also illustrates the preferred canopy positions and growth habit of the genera as well as the taxon specific trends in nitrogen isotope composition.

224 B.7. References

15 Bergerson F.J., Peoples M.B. and Turner G.L. (1990) Measurement of N2 fixation by N natural abundance in the management of legumes: Roles and precautions. In: P.M. Greshoff, L.E. Roth, G. Stacey and W.E. Newton (eds.) Nitrogen fixation: Achievements and Objectives. Proceedings of the 8th International Congress on Nitrogen Fixation, pp. 315-322. Chapman & Hall, New York.

Björkman O., Boardman N.K., Anderson J.M., Thorne S.W., Goodchild D.J. and Pyliotis N.A. (1972) Effect of light intensity during growth of Atriplex patula on the capacity of photosynthetic reactions, chloroplast components and structure. Carnegie Institue of Washington Yearbook 71, 115-135.

Coplen T.B., Brand W.A., Gehre M., Groning M., Meijer H.A.J., Toman B. and Verkouteren R.M. (2006) New guidelines for δ13C measurements. Anal. Chem. 78, 2439-2441.

Dilcher D.L., Pavlick R. and Mitchell J. (1970) Chlorophyll Derivatives in Middle Eocene sediments. Science 168, 1447-1449.

Domingues T.F., Martinelli L.A. and Ehleringer J.R. (2007) Ecophysiological traits of plant functional groups in forest and pasture ecosystems from eastern Amazônia, Brazil. Plant Ecol. 193, 101-112.

Evans J.R. and Poorter, H. (2001) Photosynthetic acclimation of plants to growth irradiance: the relative importance of specific leaf area and nitrogen partitioning in maximizing carbon gain. Plant, Cell and Env. 24, 755-767.

Evans J.R. and Terashima I. (1987) Effects of nitrogen nutrition on electron transport components and photosynthesis in spinach. Australian J. of Plant Physiol. 14, 59- 68.

Hobbie E.A., Macko S.A. and Williams M. (2000) Correlations between foliar δ15N and nitrogen concentrations may indicate plant-mycorrhizal interactions. Oecologia 122, 273-283.

Högberg P., Högberg M.N., Quist M.E., Ekblad A. and Näsholm T. (1999) Nitrogen isotope fractionation during nitrogen uptake by ectomycorrhizal and non- mycrrhizal Pinus sylvestris. New Phytologist 142, 569-576.

Marshall J.D., Brooks J.R. and Lajtha K. (2007) Sources of variation in the stable isotope composition of plants. In: R. Michener and K. Lajtha (eds.) Stable Isotopes in Ecology and Environmental Science, 2nd ed. pp. 22-60. Blackwell Publishing, Malden, MA.

225

Martinelli L.A., Piccolo M.C., Townsend A.R., Vitousek P.M., Cuevas E., McDowell W., Robertson G.P., Santos O.C. and Treseder K. (1999) Nitrogen stable isotopic composition of leaves and soil: tropical versus temperate forests. Biogeochem. 46(1-3), 45-65.

Medina E. and Minchin P. (1980) Stratification of δ13C values of leaves in Amazonian rainforests. Oecologia 45, 377-378.

Niklas K.J. and Giannasi D.E. (1978) Angiosperm paleobiochemistry of the Succor Creek flora (Miocene) Oregon, U.S.A. Am. J. Botany 65, 943-952.

Ometto J.P.F.B., Ehleringer J.R., Domingues T.F., Berry J.A., Ishida F.Y., Mazzi E., Higuchi N., Flanagan L.B., Nardoto G.B. and Martinelli L.A. (2006) The stable carbon and nitrogen isotopic composition of vegetation in tropical forests of the Amazon Basin, Brazil. Biogeochem. 79, 251-274.

Poorter H., Pepin S., Rijkers T., de Jong Y., Evans J.R. and Körner C. (2006) Construction costs, chemical composition and payback time of high- and low- irradiance leaves. J. Exp. Botany 57, 355-371.

Valladares F. and Niinemets U. (2008) Shade tolerance, a key plant feature of complex nature and consequences. Ann. Rev. Ecol. Evolution Systematics 39, 237-257.

Vitousek P.M. (1984) Litterfall, nutrient cycling, and nutrient limitation in tropical forests. Ecol. 54, 317-325.

226 Appendix C. Molecular and isotopic investigation of the Mar 2X core

C.1. Introduction

The Paleocene-Eocene Thermal Maximum (PETM; 56.3 Ma), a short-lived episode of pronounced global warming, is recorded in faunal, floral, and geochemical records globally. South American pollen records indicate that a major shift in the dominant plant families that established the first Neotropical rainforests in the western hemisphere

(Jaramillo et al., 2010) occurred prior to this interval.

Globally, a negative carbon isotope excursion (CIE) distinguishes this period (Koch et al., 1992) and is expressed in marine and terrestrial carbonates and organic matter. The

CIE is attributed to a massive release of isotopically depleted carbon into the atmosphere

(Pagani et al., 2006; Zeebe, et al., 2009) The magnitude of the CIE in terrestrial carbon differs geographically. At mid and high latitutdes, the terrestrial CIE captured in carbonates and plant lipids averages -4.6‰, whereas the CIE recorded in low-latitude areas is much smaller. Sediment core samples from Colombia (Wild Sur-1) and

Venezuela (Mar 2X and Gonzales) record a CIE of 2-3‰ (Jaramillo et al., 2010). The cause of this difference is not clear, but it has been attributed to a regional patterns in hydrology and water-use changes (Farquhar et al., 1989; Bowen et al., 2004), taxonomic shifts (Smith et al., 2007; Diefendorf et al., 2010), or forest structure and canopy effects

(Wing et al., 2005; McInerney & Wing, 2011). Building on our findings that canopy

227 cover is potentially recorded by plant-wax distributions and isotopic compositions

(chapters 3 , 4) we conducted compound-specific analysis of core sediments spanning the PETM seeking to identify the extent of canopy coverage during the PETM as well as changes in coverage before and after the event. Pollen in these cores record family composition and diversity trends similar to the Cerrejón paleoflora (Jaramillo et al.,

2010).

C.1.1. Canopy structure across the PETM

While many workers have found that forest canopies in the high latitudes became much more open during and in response to the warming associated with the PETM (Wing et al.,

2005; McInerney & Wing, 2011) there is no evidence at low latitude for a similar canopy response (Jaramillo et al., 2010). While ecosystem turnover may have occurred as a result of the warming event this does not necessarily imply changes in canopy coverage

(Burnham & Johnson, 2005). Moisture drives most of the carbon isotopic variations among biomes globally (Diefendorf et al., 2010), and this can be overprinted with even lower values when understory biomass is significantly recorded in carbon archives

(Graham et al., in review; Chapter 2). Fossil evidence suggest forests in high-latitudes experienced an opening in the forest canopy during the PETM (McInerney & Wing,

2010), and yet this is not easily reconciled with the greater magnitude of the CIE recorded at mid and higher latitudes, which potentially suggest increased moisture during seasonal plant growth. Thus the magnitude of CIE at different localities is possibly not a clear indicator of ecosystem structure changes.

228

C.1.2. Taxonomic shifts and their influence on the CIE

The palynofloral record suggests a shift from Pantropical taxa to dominance by the families currently associated with the Neotropics (Rull, 1999; Jaramillo & Dilcher, 2001;

Jaramillo, 2002). Diversity changes associated with the extirpation of certain taxa could also potentially decrease the CIE expressed in the Neotropics. In our samples from

Panama, as in many modern biomes, carbon isotopic composition can be particular to certain taxa and likely represent the water-use efficiency of the organism (Diefendorf et al., 2010; see also chapter 5). An increase in taxa that typically have greater isotope discrimination during photosynthesis (and thus more 13C-depleted foliar carbon) could have attenuated the CIE captured in bulk organic carbon. Here we examine the alkane complement in core samples from before and after both the taxonomic shift and the

PETM to evaluate potential influences of the plant community shift and climate changes, respectively, on the plant-wax record.

We extracted n-alkanes from sediment cores that span the PETM and the diversity shift.

We use plant-waxes to evaluate any changes in amounts, chain-length distribution, and

13C-content associated with the rise of Neotropical flora and abrupt climate warming.

C.2. Samples and methods

229 C.2.1. Mar 2X core

This 450 m core from the western Maracaibo basin of Venezuela (11.06º N, 72.17º W, paleolatitude = ~9.1º N) spans the PETM section of the Marcelina Formation. This formation consists of cross-bedded sandstone with interbedded shales and coal lenses.

This unit fines upward in 20 m thick intervals and is interpreted as channels in a coast or delta plain (Murat & Ruiz, 1994). This unit has been biostratigraphically dated by palynomorphs. In particular, the PETM interval of this core contains 10 key Eocene originations and 3 key Paleocene extinctions that co-occur in other formations that have been radiometrically dated by U-Pb isotopes in zircons from volcanic tuff. Further, the

CIE associated with the PETM interval can be observed in the all biostratigraphically dated cores.

C.2.2. Preparation of rock from core samples

Fifty core samples were selected that include pre- and post-CIE section. Rock samples were crushed and homogenized with a Teflon pestle in glass tubes and then treated with

6N HCl for two hours and rinsed three times with deionized water. This treatment was repeated until the final rinse water maintained a neutral pH. Samples were then lyophilized.

Carbon isotope analyses were performed on decarbonated rock samples using a continuous flow Costech elemental analyzer connected to a Thermo Finnigan Delta XP

230 IRMS. Analyses were performed under continuous flow (He; 120 ml/min), oxidized at

1020°C over chromium (III) oxide and cobaltous silver (II, III) oxide, then reduced over elemental copper at 650°C before passing through a water trap and into IRMS. δ13C values were corrected for sample size and reported relative to Vienna Pee Dee Belemnite

(VPDB) (Coplen et al., 2006). Error was determined by repeated analysis of internal standards (calibrated by IAEA CH-7, IAEA CH-6, and USGS-24; polyethylene, sucrose and graphite, respectively). Instrument precision was ±0.13‰ (n=21). Accuracy, the average difference between the measured and true δ13C value, was 0.53‰ (n = 40).

Duplicate analyses were performed for 50% of all samples. Samples that defined the end points of the isotopic range were analyzed in triplicate.

C.2.3. Alkane extraction and chromatographic preparation

Rock samples were extracted using a Dionex Accelerated Solvent Extraction system with dichloromethane/methanol (65:35, v/v). Samples were heated for 5 min at 76 bar and

100ºC, followed by a 5 min heated hold, flushed with solvent mix and purged with N2 for

150 s. n-Alkanes were chromatographically separated into polarity fractions by sequential in-cell separation by the Dionex Accelerated Solvent Extraction system as per Magill et al., in review). Total lipid extracts (TLEs) were then evaporated to near dryness under N2 before being transferred to an ashed glass fiber disc (Dionex #047017). The TLE- saturated disc was packed in a column with 1 g of high purity silica gel with a a 60 Å pore size and 70-230 mesh particle size and then 1 g of 5% silver-impregnated alumina oxide. n-Alkanes were eluted in 1:20 dichloromethane:hexanes while more polar

231 compounds were eluted with ethyl acetate, dichloromethane and methanol. Columns were heated to 50ºC for one minute at 500 psi then flush with 70% of the cell volume three times. The final elution, following the heated hold period, flushed with entire solvent mix and purged the column with N2 for 150 s.

C.2.4. n-Alkane identification, quantification and isotope analysis

Extracted alkanes were identified using a Hewlett-Packard (HP) 6890 gas chromatograph

(GC) connected to an HP 5973 quadrupole mass spectrometer (MS). We used a fused silica capillary column (Agilent J&W DB-5; 30 m, 0.32 mm, 0.25 µm) with helium carrier gas. Injections were made in hexanes and at an inlet temperature of 320ºC. The column flow rate is 2.0 ml/min and the column oven starts with an initial temperature of

60ºC for 1 min, followed by a ramp to 320ºC at 6ºC/min. This final temperature is then held for 20 min. The MS was operated with a scanning mass range of m/z 50-700 at 3 scans per second and an ionization energy of 70 eV. Compounds were identified using standards, the NIST 98 spectrum library, fragmentation patterns, the published literature, and chromatographic retention times.

Compounds were quantified on a HP 5890 GC with a flame ionization detector (FID) under the same GC conditions as for identification. Compound peak areas were normalized to co-injected internal standard containing alkane nC-16 and 1,1’-binaphthyl.

Aliquots were converted to mass quantities using response curves for 28 n-alkanes ranging from n-C12 to n-C40 analyzed in concentrations ranging from 0.2 to 200 µg/ml.

232 Concentrations were normalized to the mass of mass of total organic carbon in dry leaf, fossil or rock material extracted. Average n-alkane chain lengths are calculated using abundances with the following formula modified from Eglinton & Hamilton (1967) by

Henderson et al. (in review).

ACL29-35 = ((C29*29) + (C31*31) + (C33 *33) + (C35 *35))/(C29+C31+C33+C35)

Extracted n-alkane fractions were separated on a Varian model 3400 GC equipped with a split/splitless injector operated in splitless mode with a fused silica column (Agilent J&W

DB-5; 30 m, 0.32 mm, 0.25 µm). The GC temperature program followed the same conditions as the identification and quantification steps. Following separation, compounds were combusted over nickel and platinum wire with O2 in He (1%, v/v) at

1000ºC, with the resulting CO2 monitored using a Finnegan Mat 252. Carbon isotope values of samples are reported in delta notation relative to the VPDB scale. Within-run precision and accuracy was determined by a co-injected an internal standard containing alkanes nC-38 and nC-41. Isotopic abundances were determined relative to a reference gas calibrated with Mix A and Mix B (n-C16 to n-C30; Arndt Schimmelmann, Indiana

University). Over the course of all samples the precision and accuracy were 0.22‰ and

0.06‰ (1σ, n=146), respectively.

C.3. Results

233 Only in a few strata did the organic carbon concentrations in core samples exceed 2%

(Figure A7) and there was no appreciable change in the preservation of organic carbon during the PETM interval (as defined by the CIE; Jaramillo et al., 2001). The samples with highest %TOC were found 50 m below, during the PETM.

Alkane concentration varied widely throughout the core and was not related to the amount of organic carbon in a sample (Figure A8). The highest concentrations of alkanes were well below the CIE, in shales 60-90 m before the PETM interval. Chain-length distribution of alkanes extracted throughout the core was between 29.9. and 30.2 (Figure

A9). For comparison, the ACL29-35 of bulk rock from the Cerrejón site was between 29.0 an 29.5, however the Guaduas Formation had a lower ACL29-35 near 28.0 (see chapter 4).

There was a decrease in ACL to 29.9 just prior to the PETM however ACL29-35 varied widely, as much as 0.8 units in this section, as did the sections before and after the PETM

13 interval. The δ CTOC values for bulk rock ranged between -25.0 and -27.0‰ before the

13 PETM interval. The range in δ CTOC values became narrower after the PETM,

13 approaching -27.3‰ (Figure A9). These values are more depleted than δ CTOC measured in bulk rock from the Cerrejón site (see chapter 5).

13 The weighted mean average of alkanes (δ C25-35) extracted from cores varied little from an average of -26.0‰ (Figure A10). Prior to and just after the PETM interval there were

13 periodic δ C25-35 values nearing -24‰. The only evidence of a carbon isotope excursion in alkanes from the core were two values of -26.5‰ and -29.8‰ in the middle of the defined PETM interval.

234

C.4. Discussion

13 δ C25-35 values do not change appreciably prior to the PETM interval which would indicate that the families that become prominent on the landscape during this time are not more or less water-use efficient, which would be expressed in the carbon isotope

13 composition. The δ C25-35 values also do not change appreciably during the PETM which

13 suggests there is not a major change in precipitation regime. Alkane δ C25-35 values do not reflect an isotopic response indicative of a carbon isotope excursion in the interval where this would be expected. This would suggest that the carbon isotope excursion reflects largely non-terrestrial, or at least non-foliar carbon.

While ACL29-35 is a poor indicator of plant taxonomy (see chapter 4 and Henderson et al., in review), the difference in ACL29-35 values between the Guaduas and Cerrejón sites could be the interpreted as the result of family-level dominance patterns. If a family-level dominance pattern does affect ACL29-35 values this could be expected in the pre-PETM floral changes found in palynoflora of the Mar-2X core. This study does not find any

ACL29-35 trend accompanying these changes however, either during the period of diversity shift or during the PETM.

235 C.5. Figures

Figure C.6. %TOC from Mar 2X core samples before, during and after the PETM interval

236

Figure A7: Alkane concentrations in Mar 2X core samples normalized to the amount of organic carbon.

237

Figure C.8. Stratigraphic column for Mar 2x (from Jaramillo et al., 2010) with sample 13 sites marked and the ACL of extracted alkanes and δ CTOC values for each site. The PETM interval defined by the CIE in Jaramillo et al., 2010 is marked since the selection of core samples in our study does not define this boundary well.

238

13 Figure C.9. The weighted mean average δ C25-35 values for alkanes extracted from the Mar 2X column. Again, the samples selected capture only two points that may be interpreted as a carbon isotope excursion so the strata that define the CIE from Jaramillo et al., 2010 have been marked.

239 Appendix D. Data Tables

Table D.1. Data collected from vertical transect in Bosque Protector San Lorenzo, Panama. Environmental data was collected on site and isotopic data was collected at Pennsylvania State University. Table includes calculated photosynthetic fractionation values both for the global CO2 value (as per Scripps database) and measurements of sampled gases from within the canopy for a select number of samples.

240 241 242

243 Table D.2. Data collected from vertical transect of forest at the Smithsonian Ecological Research Center. Light data is on a relative scale and was collected via hemispheric photography. Isotopic data was generated at the Carnegie Institute of Washington.

244

245 Table D.3. Data incorporated into Monte Carlo resampling model for litter simulation and simulated leaf collection for tropical forest.

246 Table D.4. Data incorporated into Monte Carlo resampling model for litter simulation and simulated leaf collection for temperate forest.

247 Table D.5. Data collected from fossil leaves from the Guaduas Formation. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and isotopic analyses were performed at Pennsylvania State University.

248 249 Table D.6. Data collected from fossil leaves from the Cerrejón Formation, Site 315. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and isotopic analyses were performed at Pennsylvania State University.

250

.

251 Table D.7. Data collected from fossil leaves from the Cerrejón Formation, Site 318. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and isotopic analyses were performed at Pennsylvania State University.

252

253 Table D.8. Compound specific carbon isotope data collected from fossil leaves of the Guaduas Leaves were sampled from collection at the Smithsonian Tropical Research Center, and isotopic analyses were performed at Pennsylvania State University.

254 Table D.9. Compound specific carbon isotope data collected from fossil leaves of the Cerrejón Formation, Site 315. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and isotopic analyses were performed at Pennsylvania State University.

255 Table D.10. Compound specific carbon isotope data collected from fossil leaves of the Cerrejón Formation, Site 318. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and isotopic analyses were performed at Pennsylvania State University.

256 Table D.11. Compound specific carbon isotope data collected from sediment and rock samples from the Guaduas and Cerrejón Formations. Isotopic analyses were performed at Pennsylvania State University.

257 Table D.12. Accession numbers and taxonomic association of fossils that failed to yield any measurable alkanes

258 Table D.13. Compound specific carbon alkane concentration data collected from fossil leaves of the Guaduas Leaves were sampled from collection at the Smithsonian Tropical Research Center, and quantification was performed at Pennsylvania State University.

259 Table D.14. Compound specific carbon alkane concentration data collected from fossil leaves of Cerrejón Formation, Site 315. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and quantification was performed at Pennsylvania State University.

260 Table D.15. Compound specific carbon alkane concentration data collected from fossil leaves of Cerrejón Formation, Site 318. Leaves were sampled from collection at the Smithsonian Tropical Research Center, and quantification was performed at Pennsylvania State University.

261 Table D.16. Compound specific carbon alkane concentration data collected from sediment and rock samples from the Guaduas and Cerrejón Formation. Quantification was performed at Pennsylvania State University.

262 Table D.17. Lipid biosynthetic fractionation values for leaves from the fossil assemblage 13 13 (Cerrejón and Guaduas Formations) calculated using the δ Cleaf and δ C29-35 values measured at Pennsylvania State University.

263

13 Table D.18. Alkane abundances and δ Clipid values were combined in a mixing model that included the percent contribution of the morphotype to the fossils in the assemblage. Leaf census data are from Wing et al., 2009. Alkane quantification and isotope measurements were made at Pennsylvania State University. These results are the 13 predicted alkane relative abundances and δ Clipid values for bulk rock from Cerrejón Sites 315 and 318.

264 265 266

267 Table D.19. Compound specific carbon isotope data collected from San Lorenzo forest leaf samples. Isotopic analyses were performed at Pennsylvania State University.

268

269 Table D.20. Alkane abundance data collected from San Lorenzo forest leaf samples and normalized to dry leaf weight. Quantification analyses were performed at Pennsylvania State University.

270

271 Table D.21. Lipid biosynthetic fractionation values for leaves from San Lorenzo forest. 13 13 εlipid values were calculated using the δ Cleaf and δ C29-35 measurements made at Pennsylvania State University and reported on the previous pages.

272

273 Table D.22. Alkane abundance and chain-length distribution data taken from the literature for taxa ranging from Magnoliids to core eudictos. ACL is calculated using abundance values only for chain-lengths C29 – C35 as is recommended in Henderson et al., (in review).

274

275

276 277 278 279 280

281 13 Table D.23. Total organic carbon and δ CTOC values for samples from the Mar-2x core. All measurements were made at Pennsylvania State University.

282

283 13 Table D.24. Compound specific δ Clipid values for alkanes extracted from selected samples taken from the Mar-2x core. All measurements were made at Pennsylvania State University.

284

285 Table D.2. Abundance data for all alkanes extracted from selected samples of the the Mar-2x core. All measurements were made at Pennsylvania State University.

286

287

Heather V. Graham 1612 North Calvert Street, Baltimore, MD 21202 (410) 949-5193 [email protected]

Education A.B., cum laude, Chemistry, Occidental College 2007

Appointments & Awards Committee on Institutional Cooperation/Smithsonian Fellowship Award 2012 FameLab Finalist 2012 Richard R. Parizek Graduate Fellowship 2011 Pennsylvania Space Grant Graduate Student Fellowship 2010 – 2012 CarbonEARTH NSF GK12 Education Fellow 2010 – 2012 Shell Geosciences Energy Research Facilities Award 2010 – 2012 Smithsonian Tropical Research Institute Research Fellowship 2010 Teaching Assistant, Earth Materials, The Pennsylvania State University 2009 Charles E. Knopf Memorial Scholarship 2008 Ann C. Wilson Graduate Student Research Award 2008 – 2009 University Graduate Fellowship 2008 – 2009 Owen Scholars Award, Johns Hopkins University 2007 – 2008 Chemistry Lab Teaching Assistant 2006 – 2007 Rose Hills Foundation Scholarship Recipient 2006 Lawrence Berkeley National Laboratory Undergraduate Research Fellow 2004 – 2006 Santa Monica College Deans List 2002 – 2004

Publications Graham, H.V., Quinn, N.W.T. 2005 A comparative study of three soil salinity determination protocols. Journal of Undergraduate Research 5 pp. 177 Graham, H.V., Quinn, N.W.T. 2006 Evaluating the effects of tailwater irrigation on soil salinity and discharge water quality. Journal of Undergraduate Research 6 pp. 182 Jahren, A.H., Byrne, M.C., Graham, H.V., Sternberg, L.S.L., Summons, R.E. 2009 The environmental water of the middle Eocene Arctic: Evidence from δD, δ18O and δ13C within specific compounds. Palaeogeography, Palaeoclimatology, Palaeoecology 271(1-2): 96 Diefendorf, A.F., Freeman, K.H., Wing, S.W., Graham, H.V. 2011 Production of n-alkyl lipids in living plants and implications for the geologic past. Geochimica et Cosmochimica Acta 75(23):7472 Kennett, D.J., Hajdas I., Culleton, B.J., Belmecheri S., Martin S., Neff, H., Awe J., Graham, H.V., Freeman K.H., Newsom L., Lentz, D.L., Anselmetti F.S., Robinson M., Marwan, N., Southon J., Hodell, D.A., Haug, G.H. 2013 Correlating the ancient Maya and modern European calendars with high-precision AMS 14C Dating. Nature Scientific Reports 3:1597 Graham, H.V., Church, E.D., House, C.H. 2013 Packing for another Planet: Learning scientific methodology through alternative reality gaming In: Martin C., Ochsner A., Squire K. (eds.) Proceedings of the 9th Games, Learning and Society Conference (forthcoming). ETC Press.