https://doi.org/10.1130/G46152.1 Manuscript received 18 November 2018 Revised manuscript received 3 July 2019 Manuscript accepted 29 July 2019 © 2019 Geological Society of America. For permission to copy, contact [email protected]. Published online XX Month 2019 Canopy structure in Late Cretaceous and Paleocene forests as reconstructed from carbon isotope analyses of fossil leaves Heather V. Graham1,2, Fabiany Herrera3, Carlos Jaramillo4,5, Scott L. Wing6 and Katherine H. Freeman2 1NASA Goddard Spaceflight Center, Code 691, Greenbelt, Maryland 20771, USA 2Department of Geosciences, The Pennsylvania State University, University Park, Pennsylvania 16802, USA 3Chicago Botanic Garden, 1000 Lake Cook Road, Glencoe, Illinois 60022, USA 4Smithsonian Tropical Research Institute, Box 0843-03092, Balboa, Panama 5ISEM, Université de Montpellier, CNRS, EPHE, IRD, 34090 Montpellier, France 6Department of Paleobiology, National Museum of Natural History, Smithsonian Institution, P.O. Box 37012, Washington, D.C. 20013, USA ABSTRACT observed that the stable carbon isotope com- 13 While modern forests have their origin in the diversification and expansion of angiosperms position of leaves (δ Cleaf) declines strongly in the Late Cretaceous and early Cenozoic, it is unclear whether the rise of closed-canopy downward from upper canopy to understory tropical rainforests preceded or followed the end-Cretaceous extinction. The “canopy effect” (Vogel, 1978). This “canopy effect” provides is a strong vertical gradient in the carbon isotope (δ13C) composition of leaves in modern a promising approach that could be applied to closed-canopy forests that could serve as a proxy signature for canopy structure in ancient relatively common leaf compression fossils. If forests. To test this, we report measurements of the carbon isotope composition of nearly this isotope gradient is preserved in fossils, it 200 fossil angiosperm leaves from two localities in the Paleocene Cerrejón Formation and would allow canopy placement to be estimated one locality in the Maastrichtian Guaduas Formation of Colombia. Leaves from one Cer- for fossil leaves and leaf fragments. rejón fossil assemblage deposited in a small fluvial channel exhibited a 6.3‰ range in δ13C, Three major mechanisms contribute to the consistent with a closed-canopy forest. Carbon isotope values from lacustrine sediments in canopy effect. (1) High rates of respiration by the Cerrejón Formation had a range of 3.3‰, consistent with vegetation along a lake edge. soil biota combined with restricted atmospher- 13 An even-narrower range of δ C values (2.7‰) was observed for a leaf assemblage recov- ic mixing create elevated CO2 concentrations 13 13 ered from the Cretaceous Guaduas Formation, and suggests vegetation with an open canopy and C-depleted CO2 (δ Catm) in the understo- structure. Carbon isotope fractionation by Late Cretaceous and early Paleogene leaves was ry (Brooks et al., 1997; Medina and Minchin, in all cases similar to that by modern relatives, consistent with estimates of low atmospheric 1980). (2) Higher humidity lower in the under- CO2 during this time period. This study confirms other lines of evidence suggesting that story permits stomata to remain open without closed-canopy forests in tropical South America existed by the late Paleocene, and fails to loss of leaf water, resulting in a fuller expression find isotopic evidence for a closed-canopy forest in the Cretaceous. of 13C fractionation during enzymatic carbon fixation (Δleaf; Ehleringer et al., 1986; Madha- INTRODUCTION developed, and estimates of their origin range van et al., 1991). (3) High light in the upper Closed-canopy tropical forests are the most from the mid-Cretaceous to the early Paleogene canopy increases the rate of photosynthesis up diverse modern biome and can drive water, (Burnham and Johnson, 2004). Time-calibrated to four times that of leaves in the understory, and carbon, and climate dynamics at continen- molecular phylogenetic trees constructed for leads to less 13C discrimination (Zimmerman and tal and global scales (Burnham and Johnson, extant angiosperms place the modern tropi- Ehleringer, 1990; Hanba et al., 1997). As a result 2004). Although tropical rainforests comprise cal rainforest lineages as far back as 100 Ma of these pronounced gradients in CO2, water, and 13 only ∼12% of Earth’s surface, they account and could indicate that angiosperm-dominated, light, closed-canopy forest δ Cleaf values range for ∼45% of the carbon in terrestrial biomass closed-canopy forests have been present since as much as 10‰ from the sun-lit canopy top to (Watson et al., 2000; Malhi et al., 2002). These the mid-Cretaceous (Soltis and Soltis, 2004; the dark and humid understory. forests help maintain consistent temperatures Davis et al., 2005), except that fossils docu- A Monte Carlo–style leaf resampling mod- and the wet conditions (mean annual precipita- menting the morphological and ecological traits el from closed-canopy forest litter has shown 13 tion ≥2000 mm/yr) to which they are adapted common to canopy-forming angiosperms are that the wide diagnostic range of δ Cleaf values via their low albedo and massive movement rare until the Paleocene (Bruun and Ten Brink, unique to the closed-canopy forest can be found of transpired water across continents, both of 2008; Herrera et al., 2014). Further, leaf features by carbon isotope measurements from as few as which influence large-scale atmospheric circu- that indicate dense canopy can reflect multiple 50 leaves (Graham et al., 2014). Here we use 13 lation and temperatures (Bastable et al., 1993; drivers, leaving few empirical tools that can be δ Cleaf to estimate canopy structure in leaf fos- Betts, 1999; Bonan, 2008; Boyce et al., 2010). used to assess ancient forest structure (Beerling sil assemblages from the Maastrichtian Guaduas It is not well understood when angiosperm- and Royer, 2002; Feild et al., 2011; Carins Mur- Formation and Paleocene Cerrejón Formation 13 dominated closed-canopy tropical forests first phy et al., 2014). In modern forests, it has been of Colombia. We also use fossil δ Cleaf data in CITATION: Graham, H.V., et al., 2019, Canopy structure in Late Cretaceous and Paleocene forests as reconstructed from carbon isotope analyses of fossil leaves: Geology, v. 47, p. 977–981, https://doi.org/10.1130/G46152.1 Geological Society of America | GEOLOGY | Volume 47 | Number XX | www.gsapubs.org 1 Downloaded from https://pubs.geoscienceworld.org/gsa/geology/article-pdf/4825714/g46152.pdf by Northwestern University user on 09 September 2019 combination with the predicted δ13C values of paleoatmospheric CO to determine if photosyn- Figure 1. Photos of mega- 2 fossils representing thetic fractionation (Δleaf) differs greatly between preservation of sampled these leaves and their modern descendants. specimens (see Wing 13 δ Cleaf values reflect the source CO2 composi- et al. [2009] for Cerrejón Formation, Columbia, tion as well as the carbon isotope discrimina- A B C tion occurring during photosynthesis. Fraction- fossils) from all three fossil localities, Cerre- ation is subject to environmental influences and D E jón (sites 0318 and 0315) genetic factors that affect isotopic expression and Guaduas Formations, trends (Hubick et al., 1990). Comparison of Colombia. Morphotype the Δ values from modern plants with those identifier and best-guess leaf family are given for each. of their fossil ancestors would indicate how Scale bars: 5 cm. (A) conserved these fractionation trends are within Euphorbiaceae, CJ10. (B) plant families. Menispermaceae, Meni- spermites cerrejonensis, MATERIALS AND METHODS CJ6. (C) Sapotaceae, CJ8. (D) Sapotaceae, CJ8. (E) 13 This study compares δ Cleaf from three fos- Fabaceae, CJ1. (F) Inde- sil assemblages with that of leaves from mod- terminate dicot. (G) Dicot, 13 F G GD14. ern forests to determine if the δ Cleaf range pre- served in the fossil cuticles is consistent with a closed canopy (see representative specimens in Fig. 1). All three fossil assemblages were were associated with ten morphotypes from nine indicate an estimated MAT of 22.1 ± 3.4°C collected in Colombia (Fig. 2): two from the families, as well as a selection of taxonomically and MAP of ∼2.4 m/yr (Gutierrez and Jara- Paleocene Cerrejón Formation and one from indeterminate non-monocot (magnoliid or eu- millo, 2007). At the time of deposition, this the Late Cretaceous (Maastrichtian) Guaduas dicot) angiosperm leaves. At site 0318, leaves location was a coastal plain similar to that of Formation. Both Cerrejón assemblages include were collected from a laterally extensive, thinly the Cerrejón Formation (Gutierrez and Jara- many of the families dominant in modern bedded, flat-laminated siltstone interpreted as a millo, 2007). Fossil leaves from the Guaduas closed-canopy forests of the Neotropics (e.g., shallow lake deposit. Sampled leaves included Formation were taken from laminated and Fabaceae, Arecaceae, Lauraceae), and physi- 19 morphotypes from 10 families as well as a massive mudstones with sandstone interbeds ognomic leaf features—size, entire margins, selection of indeterminate non-monocot leaves. above fine-grained sand beds intercalated with vein density—that indicate a closed-canopy, Herrera et al. (2008) and Wing et al. (2009) de- coal seams (Guerrero, 2002). Most angiosperm multi-layered rainforest (Wing et al., 2009; scribed family identification and morphotype leaves from the Guaduas flora can only be de- Herrera et al., 2011). In contrast, the Guaduas assignment. scribed as indeterminate dicots coexisting with Formation paleoflora neither is physiognomi- Late Cretaceous fossils were collected from abundant gymnosperms in a community that cally similar to contemporary closed-canopy the middle Guaduas Formation of Boyacá De- has no modern analog. One leaf for this study communities nor includes taxa assigned to partment (5°55′45′′N, 72°47′43′′W). Palyno- could be assigned to a family, and nine others extant families dominant in Neotropical rain- flora indicate an age of ca. 68–66 Ma (Muller could be assigned to one of five morphotypes forests (Guierrez and Jaramillo, 2007). These et al., 1987).