Leaf Wax N-Alkane Distributions in and Across Modern Plants: Implications for Paleoecology and Chemotaxonomy

Available online at www.sciencedirect.com

Geochimica et Cosmochimica Acta 117 (2013) 161–179 www.elsevier.com/locate/gca

Leaf wax n-alkane distributions in and across modern plants: Implications for paleoecology and chemotaxonomy

Rosemary T. Bush a,⇑, Francesca A. McInerney a,b

a Department of Earth and Planetary Sciences, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208-3130, USA b Sprigg Geobiology Centre, Environment Institute and School of Earth and Environmental Sciences, University of Adelaide, Mawson Laboratories, Adelaide, South Australia 5005, Australia

Received 20 June 2012; accepted in revised form 18 April 2013; available online 29 April 2013

Abstract

Long chain (C21 to C37) n-alkanes are among the most long-lived and widely utilized terrestrial plant biomarkers. Dozens of studies have examined the range and variation of n-alkane chain-length abundances in modern plants from around the world, and n-alkane distributions have been used for a variety of purposes in paleoclimatology and paleoecology as well as chemotaxonomy. However, most of the paleoecological applications of n-alkane distributions have been based on a narrow set of modern data that cannot address intra- and inter-plant variability. Here, we present the results of a study using trees from near Chicago, IL, USA, as well as a meta-analysis of published data on modern plant n-alkane distributions. First, we test the conformity of n-alkane distributions in mature leaves across the canopy of 38 individual plants from 24 species as well as across a single growing season and find no significant differences for either canopy position or time of leaf collection. Sec- ond, we compile 2093 observations from 86 sources, including the new data here, to examine the generalities of n-alkane parameters such as carbon preference index (CPI), average chain length (ACL), and chain-length ratios for different plant groups. We show that angiosperms generally produce more n-alkanes than do gymnosperms, supporting previous observations, and furthermore that CPI values show such variation in modern plants that it is prudent to discard the use of CPI as a quantitative indicator of n-alkane degradation in sediments. We also test the hypotheses that certain n-alkane chain lengths predominate in and therefore can be representative of particular plant groups, namely, C23 and C25 in Sphagnum mosses, C27 and C29 in woody plants, and C31 in graminoids (grasses). We find that chain-length distributions are highly variable within plant groups, such that chemotaxonomic distinctions between grasses and woody plants are difficult to make based on n-alkane abundances. In contrast, Sphagnum mosses are marked by their predominance of C23 and C25, chain lengths which are largely absent in terrestrial vascular plants. The results here support the use of C23 as a robust proxy for Sphagnum mosses in paleoecological studies, but not the use of C27,C29, and C31 to separate graminoids and woody plants from one another, as both groups produce highly variable but significant amounts of all three chain lengths. In Africa, C33 and C35 chain lengths appear to distinguish graminoids from some woody plants, but this may be a reflection of the differences in rainforest and savanna environments. Indeed, variation in the abundances of long n-alkane chain lengths may be responding in part to local environmental conditions, and this calls for a more directed examination of the effects of temperature and aridity on plant n-alkane distributions in natural environments. Ó 2013 Elsevier Ltd. All rights reserved.

1. INTRODUCTION

Long chain normal alkanes (n-alkanes), C21–C37, are ⇑ Corresponding author. Tel.: +1 832 588 7290; fax: +1 847 491 synthesized as part of the epicuticular leaf wax of terrestrial 8060. plants and are among the most recognizable and widely E-mail address: [email protected] (R.T. Bush). used plant biomarkers, with a history of research stretching

0016-7037/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.gca.2013.04.016 162 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 back almost a century (Chibnall et al., 1934). Plants typi- have contributed to fossilized soil and sedimentary organic cally produce a range of n-alkanes, commonly with a strong matter. Understanding intra-plant variability in n-alkane odd-over-even predominance and one or two dominant production is critical to applying n-alkanes to paleoecolog- chain lengths (Eglinton and Hamilton, 1963, 1967). Because ical interpretations, as it is often difficult to assess the can- they are straight-chain hydrocarbons lacking functional opy position and age of leaves that have contributed to groups, n-alkanes are especially stable and long-lived mole- sedimentary n-alkane records. cules that can survive in the fossil record for tens of millions A great deal of research has been devoted to identifying, of years (Eglinton and Logan, 1991; Peters et al., 2005). n- quantifying, and interpreting naturally occurring leaf wax Alkanes occur in both modern and fossil leaves (e.g. Huang n-alkanes in modern plants, often with the goal of using et al., 1995; Otto et al., 2005); in soils, paleosols, and fluvial them as taxon-specific chemical fingerprints. Chemotaxo- sediments (e.g. Quenea et al., 2004; Smith et al., 2007); and nomic studies often use one or a few plants from a single in both lacustrine and marine sediments (Schefuß et al., location to represent a species (e.g. Maffei et al., 2004); 2003; Sachse et al., 2004; Handley et al., 2008). The stable however, n-alkane distributions can vary within a species isotopic compositions (d13C and dD) of n-alkanes and their across its range (Dodd and Afzal-Rafii, 2000; Dodd and applications in paleoecology and paleoclimatology have Poveda, 2003). Geochemical studies of sedimentary long- been studied extensively (see Castan˜eda and Schouten chain n-alkanes have focused on the application of ratios (2011) and Sachse et al. (2012) for review). Long chain n-al- of particular chain lengths in an effort to reconstruct past kanes have great potential to inform us on past terrestrial ecosystems. The ratio of longer chain lengths (e.g. C29 or ecosystems and environments, but their interpretation as C31)toC17 has been used as a proxy for relative inputs of paleo-proxies requires a strong understanding of variations terrestrial plants versus aquatic algae and phytoplankton in n-alkane production both within and between modern in lake sediments (Cranwell et al., 1987; Meyers and Ishiwa- plants. tari, 1993). n-Alkanes with more intermediate chain n-Alkanes contribute to the hydrophobic properties of lengths—C23, and to a lesser extent C25—have been utilized leaf wax and serve as part of the plant’s first barrier from to model Sphagnum peat moss (Nott et al., 2000; Pancost the external environment, protecting the leaf from water et al., 2002) and aquatic plants (Ficken et al., 2000; Mu¨gler loss via evaporation (Post-Beittenmiller, 1996; Jetter et al., 2008). Longer chain lengths have also been used as et al., 2006). Early studies suggest that leaf wax n-alkanes identifiers for different vascular plant groups. For example, accumulate rapidly during leaf maturation in spring and in studies of lake sediments it has been postulated that C31 early summer, and that hydrocarbon amounts remain rela- represents input from grasses while C27 and C29 represent tively constant through the remainder of the growing sea- input from trees and shrubs (Meyers and Ishiwatari, 1993; son (Eglinton and Hamilton, 1967; Avato et al., 1984; Meyers, 2003). This reasoning has been used to interpret ra- Gu¨lz and Mu¨ller, 1992; Tipple et al., 2013), although this tios constructed from these n-alkanes in lake cores and loess may not necessarily mean that n-alkane proportions are sequences as changes in ecosystem structure over time cor- maintained (Strańsky´ et al., 1967). Sachse et al. (2010) responding with climate or land use change (Brincat et al., and Tipple et al. (2013) demonstrate that isotopic values 2000; Schwark et al., 2002; Hanisch et al., 2003; Zhang of leaf wax n-alkanes reflect the isotopic values of leaf water et al., 2006). These n-alkane ratios often can be correlated at the time of leaf formation, but other studies found that with other evidence of plant community change, such as the isotopic composition of leaf wax n-alkanes changes pollen records, and are compelling in such context. How- through the growing season (Lockheart et al., 1997; ever, the supposition that single n-alkanes represent such Chikaraishi et al., 2004; Sachse et al., 2009), suggesting that large plant groups as grasses and woody plants is based turnover of surface leaf wax may depend on plant type or on relatively sparse original data (Wakeham, 1976; environmental conditions (e.g. wind). The rate of wax turn- Cranwell, 1984; Kawamura and Ishiwatari, 1984; Cranwell over and variations in hydrocarbon production across spe- et al., 1987), and although widely cited, Cranwell (1973) cies due to environmental pressures, e.g. temperature, does not present any data on the distribution of n-alkanes aridity, or ablation by wind, are relatively unknown (Jetter in grasses in contrast to trees. et al., 2006). Furthermore, within an individual tree, leaf In addition to ratios of individual n-alkane abundances, physiology can vary from sun-exposed to shaded canopy several other methods for characterizing a given n-alkane positions, and spatial variation in overall n-alkane produc- distribution have been developed. The two most common tion across the canopy has been observed for some tree spe- are average chain length (ACL) and carbon preference in- cies (Dyson and Herbin, 1968; Lockheart et al., 1997). dex (CPI). ACL is the weighted average of the various car- Similarly, variation in n-alkane distribution across a tree’s bon chain lengths, usually defined as canopy or across a growing season could undermine the ACL ¼ RðC nÞ=RðC Þ; power of comparisons between modern leaves, typically n n collected during the height of the growing season, and sed- where Cn is the concentration of each n-alkane with n car- imentary n-alkane profiles derived from leaves that were bon atoms. Carbon preference index measures the relative likely, although not necessarily, naturally abscised. Most abundance of odd over even carbon chain lengths, where leaves in the fossil record and the leaves which would have CPI ¼½RoddðC21–33ÞþRoddðC23–35Þ=ð2RevenC22–34Þ; contributed n-alkanes to the sedimentary record likely grew on the part of the canopy with the greatest amount of light and captures the degree to which odd carbon number exposure (Greenwood, 1991), but shaded leaves also would n-alkanes dominate over even carbon numbers R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 163

(see Marzi et al., 1993). CPI values greater than 1 mean a Table 1 predominance of odd over even chain lengths. In sediments, Species collected for this study from the Lincoln Conservatory CPI > 1 is used to indicate a terrestrial plant source and (LC), Chicago, IL and from the Chicago Botanic Garden (CBG), thermal immaturity of the source rock (Bray and Evans, Glencoe, IL. Species marked in bold were collected from green- 1961; Eglinton and Hamilton, 1967). Other numerical house plants; all other plants were grown outdoors. parameters have also been used to describe and distinguish Species Family Location n-alkane distributions, including the location-specific ma- Pteridophytes trix model developed by Jansen and coworkers (Jansen Cyathea cooperi Cyatheaceae CBG et al., 2006, 2010). Matteuccia pensylvanica Polypodiaceae CBG The robust application of n-alkane distributions as Osmunda regalis Osmundaceae CBG paleoecological biomarkers requires the systematic survey Gymnosperms of variation among different modern plants (Diefendorf Cycas circinalis Cycadaceae CBG et al., 2011). With the wealth of studies now published, a Ginkgo biloba Ginkgoaceae CBG more comprehensive analysis of n-alkanes in modern plants Larix decidua Pinaceae CBG is now possible and can inform their use as paleoecological Metasequoia glyptostroboides Cupressaceae CBG indicators. Therefore, this study uses a combination of new Picea abies Pinaceae CBG data on n-alkane distributions from plants growing at the Pinus sylvestris Pinaceae CBG Chicago Botanic Garden and the Lincoln Conservatory Taxodium distichum Cupressaceae CBG Taxus cuspidata Taxaceae CBG (Chicago, IL, USA) as well as a survey of the published Thuja occidentalis Cupressaceae CBG n-alkane literature (totaling 2093 n-alkane measurements from 86 sources) to examine the following sets of questions: Angiosperms Alnus glutinosa Betulaceae CBG Annona muricata Annonaceae LC (1) Do n-alkane distributions vary within a plant with Artocarpus altilis Moraceae LC canopy position or leaf age? To test this, leaves were Asimina triloba Annonaceae CBG simultaneously collected from sun-exposed and Carpinus caroliniana Betulaceae CBG shaded portions of a tree canopy during the summer Celtis occidentalis Ulmaceae CBG and autumn across 38 individual plants: 22 different Fagus sylvatica Fagaceae CBG outdoor angiosperm and gymnosperm tree species Gleditsia triacanthos Fabaceae CBG and two ferns from the Chicago Botanic Garden. Koelreuteria paniculata Sapindaceae CBG (2) How do n-alkane chain-length distributions vary Lindera benzoin Lauraceae CBG among plant types? How does the absolute abun- Platanus orientalis Platanaceae CBG dance of n-alkanes differ between angiosperms and Populus deltoides Salicaceae CBG Pterocarya stenoptera Juglandaceae CBG gymnosperms? Compiled literature data on n-alkane Rhus typhina Anacardiaceae CBG distributions and abundances from locations on Salix alba Salicaceae CBG every continent except Antarctica, as well as new Tamarindus indica Fabaceae CBG measurements from the Chicago Botanic Garden Tilia cordata Tiliaceae CBG and Lincoln Conservatory, are analyzed to address Washingtonia robusta Arecaceae CBG these questions. (3) Can the abundance of different chain lengths be used to reconstruct plant types? How well do C23 and C25, similar climates, with windows open to the outside for air C27 and C29, and C31 actually distinguish Sphagnum circulation and daytime temperatures 27–32 °C in sum- mosses, woody plants, and grasses, respectively? mer. One to several leaves (depending on leaf size) were col- These questions were similarly addressed using the lected from at least one individual per species, and five compilation of new and published data. individuals each for the species Alnus glutinosa, Gleditsia triacanthos, Pinus sylvestris, and Taxodium distichum. To en- The purpose of this work is to address questions con- sure equal and full light exposure, only isolated trees that cerning the use of n-alkane chain lengths as a proxy for stood free from surrounding trees were selected when possi- plant types, and the answers to these questions are funda- ble. ‘Sun’ leaves were gathered from the outer edge of the mental to paleoecological interpretations of sedimentary canopy, on the southern side of each plant, and typically n-alkane distributions. at the maximum extent of the pole pruners (3 m height). ‘Shade’ leaves were taken from the bottom, northern side, 2. METHODS and as near to the trunk as possible. Leaf samples were collected in both the summer (July 2007) and fall (October– 2.1. Sample species and collection November 2007). Fall samples were collected as close to the time of abscission from the tree as possible, although Leaf samples were collected from the grounds and trop- for a subset of species, multiple leaf samples were col- ical greenhouse of the Chicago Botanic Garden (CBG) in lected—either at two different dates in the fall or of two or Glencoe, IL, and from the tropical greenhouse at the Lin- more colors of foliage representing differing levels of leaf de- coln Conservatory (LC), in Lincoln Park, Chicago (Table 1). cay on the tree (green, yellow, or brown). Shade leaves were Chicago Botanic Garden and Lincoln Park greenhouses had not collected for all outdoor species in fall because with 164 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 attenuating canopy cover, shadiness of samples could not be may not be true in all cases. (Note that data from outdoor verified. Each leaf sample was stored in a paper bag and al- plants at the Chicago Botanic Garden included in the lit- lowed to dry completely before further preparation. erature compilation are for summer, sun leaf samples only.) All statistical analyses were performed with Stata 2.2. n-Alkane identification and quantification IC 11.1.

Leaves from coniferous species were separated and 3. RESULTS stored in clean glass jars by age cohort. Evergreen and deciduous leaves of the same age were compared to one an- 3.1. Average chain length other, to avoid effects of age variability (Brooks et al., 1997). Dried leaves were homogenized by grinding in liquid Using both the new data from the Chicago Botanic Gar- N2 with mortar and pestle. Lipids were extracted from 0.2– den and Lincoln Conservatory as well as the collected liter- 0.4 g of leaf sample in 20–30 ml dichloromethane:methanol ature data, average chain length was calculated twice: first (9:1 v/v) using a Microwave Accelerated Extraction System using odd chain lengths only (ACLodd, n = 1964), and then (MarsX) with a ramp to 100 °C over 5 min, a hold time at again using both odd and even chain lengths (ACLall, 100 °C for 15 min, and cooling time of 30 min. Total lipid n = 1736). The two calculated ACL values were then re- extracts were concentrated under a stream of dry N2. gressed across all data and found to be strongly and signif- Non-polar lipids, including n-alkanes, were separated from icantly correlated (linear regression, r2 = 0.937, the polar lipids via short column silica gel chromatography p < 0.00005, Fig. A-1). In a Student’s t-test, the two groups (Eglinton and Hamilton, 1967), using 1 g of activated sil- are significantly different (p < 0.00005), but the difference is ica gel in a Pasteur pipette plugged with glass wool and small (0.166). And when the cycad family Zamiaceae 4 mL hexanes. Once isolated, non-polar lipids were again (Osborne et al., 1989, 1993) is excluded, the correlation concentrated by evaporation with dry N2 and analyzed by becomes slightly stronger (linear regression, n = 1654, gas chromatography/mass spectrometery (GC–MS) r2 = 0.939, p < 0.00005) and the difference between the (Medeiros and Simoneit, 2007). Samples were passed means smaller (0.159). Because 15 among 86 of the litera- through the gas chromatograph (Thermo Scientific Trace ture sources published distribution values for odd chain GC Ultra, with 15 m, 0.25 mm ID Thermo TR-5ms SQC lengths only, ACLodd was used in all subsequent analyses column) for separation and then to a quadrapole mass spec- and is referred to only as ACL in subsequent text. trometer for identification (Thermo Scientific DSQII). Sam- ples were simultaneously injected into a separate column 3.2. n-Alkane distributions within trees (with the same specifications) in the same instrument and analyzed by a flame ionization detector (FID) for assess- There was no significant difference in ACL values be- ment of relative abundance. The GC oven temperature tween sun and shade leaves of outdoor plants from the Chi- method is as follows: initial temperature at 100 °C, hold cago Botanic Garden, controlling for collection date, in for 2 min, then ramp at 11 °C/min to 320 °C, and hold at either angiosperms or gymnosperms in a paired Student’s 320 °C for 5 min. All samples were compared to a standard t-test (two-tailed, p = 0.2836). In linear regression analysis, homologous series of n-alkanes from C21 to C40, inclusive sun and shade ACL values were strongly correlated in both (Fluka, Sigma–Aldrich). angiosperms (r2 = 0.914, p < 0.00005) and gymnosperms (both evergreen and deciduous, r2 = 0.742, p < 0.00005, 2.3. Literature analysis Fig. 1A). For one species of gymnosperm, T. distichum, sun leaves had higher ACL values than shade leaves, but Data for comparisons of n-alkane distributions among with a sample size of only 3, the results were not statistically plant groups were collected from the published literature, significant; other deciduous gymnosperms do not show the which reported either tables or graphs of n-alkane same pattern as T. distichum. Similarly, there was no signif- amounts—either relative (% of total) or absolute (mass icant difference in ACL values between summer and fall unit) amounts, both for odd and even chain lengths or so- leaves, controlling for canopy position, in both angio- lely odd chain lengths. Publications reporting only derived sperms and gymnosperms in paired Student’s t-test (two- parameters such as n-alkane ratios, ACL, or CPI, without tailed, p = 0.6923). Summer and fall ACL values were reporting chain-length abundances or amounts, were not strongly correlated in linear regression in angiosperms included in the analysis. The total number of observations (r2 = 0.976, p < 0.00005) and gymnosperms (r2 = 0.817, compiled was 2093. A supplementary spreadsheet is pro- p < 0.00005, Fig. 1B). In Thuja occidentalis, even as sun vided with all literature data and their source references leaves go from green to yellow to brown in the fall, shortly as well as all new data from the Chicago Botanic Garden preceding abscission, ACL values remained constant and Lincoln Conservatory. The spreadsheet includes only (ACL = 34.53, 34.54, and 34.42 for green, yellow, and the chain lengths that were reported in the source publica- brown leaf samples, respectively). tions, with unreported chain lengths left blank; in order to assimilate diverse data, subsequent analyses assume that 3.3. n-Alkanes across plant functional types unreported chain lengths reflect an absence of that chain length (i.e. calculations assume 0% C35 where the source Plants were assigned categories based on phylogenetic do- reported abundances for C27–C33 only), although this main and growth habit: mosses, ferns, woody gymnosperms R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 165

Fig. 1. (A) Average chain length (ACL) of sun leaves versus shade leaves collected on the same day within the canopy of angiosperm trees (blue diamonds) and gymnosperm trees (ﬁlled red circles are evergreen, open circles deciduous). (B) ACL of summer leaves versus fall leaves collected at the same tree canopy location, as well as summer and fall ACL values for two outdoor ferns (green crosses). Dashed lines represent 1:1 lines. All data from the Chicago Botanic Garden. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

(conifers, cycads, and ginkgo), woody angiosperms, forbs Similarly to ACL, there is a large range of CPI values (herbaceous, non-woody dicots), graminoids (grasses, sedges, across plant groups (Fig. 2B). Of 1722 total observations, and rushes), aquatic plants, and succulents. Fig. 2 shows the the highest CPI value is 99 (woody angiosperms, Vellozia- distributions of ACL, CPI, and selected n-alkane ratios across ceae), while the lowest CPI value is 0.039 (graminoid, Poa- these groups. (Note that Fig. 2D–H report moss values for ceae). 96.0% of all CPI values are greater than or equal to 1; Sphagnum only, excluding other moss species, as discussed be- 81.2% greater than or equal to 2; and 60.7% greater than or low.) Fig. 3 shows ACL distributions for woody angiosperms equal to 5. Median CPI for all observations is 7.06; mean and gymnosperms from temperate latitudes (40°–60°)sepa- CPI is 10.69. Lastly, the large number of outlier CPI values rated into evergreen and deciduous taxa. The evergreen/decid- for some plant groups demonstrates the large range in CPI uous comparison is limited to temperate latitudes where leaf values and the asymmetrical group skew towards relatively deciduousness can be defined based on whether leaves lower values for all groups and across plant families. Dom- over-winter on the plant or are dropped in autumn, as decid- inance, or the relative percent abundance of the most dom- uousness in this region is largely based on temperature and inant n-alkane in each distribution, also has a large range, light-based winter seasonality. For tropical and sub-tropical from 100% (woody gymnosperms, woody angiosperms, plants, assessing a binary deciduous/evergreen trait is more succulents, multiple families) to 15.7% (succulents, Cacta- difficult because deciduousness can be driven by other factors ceae) at the lowest (Fig. 2C). (e.g. drought) and leaf life span exhibits more of a continuum. Measurements of absolute n-alkane amounts are re- Temperate gymnosperms show a similar distribution of ACL corded in the literature less frequently than relative values for deciduous and evergreen taxa, despite the difference amounts. Most sources report measurements as lg in sample size between the two groups (Fig. 3A). By contrast, n-alkanes/g dry leaf material. A few sources report mg in temperate woody angiosperms, evergreen species have high- n-alkanes/g wax, but this is more difficult to measure and er ACL values on average than do deciduous species, although less meaningful for applications where comparisons between both groups cover a similar range of ACL values (Fig. 3B). n-alkane amounts and biomass production are important. However, almost all of the evergreen angiosperm species (97 Angiosperms have on average 506 lg n-alkanes/g dry leaf of 101) were drawn from a single plant family, Ericaceae, (n = 282, std. dev. = 497). In contrast, gymnosperms have and thus this group may suffer from sampling bias. Overall, an average of 46 lg n-alkanes/g dry leaf material (n = 120, the distribution of ACL values is similar across all groups std. dev. = 335, Fig. 4); excluding one outlier (Podocarpus (Fig. 2A), except mosses, which have lower values. It should latifolius) with reported 3607 lg n-alkanes/g dry leaf (Ficken be noted that the relatively low ACL values in mosses are et al., 2000), the average for gymnosperms drops to 16 lg n- found predominantly in Sphagnum species, and not other alkanes/g dry leaf (n = 119, std. dev. = 69). n-Alkane quan- moss genera (Nichols et al., 2006). The group with the smallest tities for only three Sphagnum moss species, reported as lg range of values, aquatic plants, is also the group with the n-alkanes/g dry plant material, were included in this study smallest number of data; all other groups demonstrate a large (Pancost et al., 2002), and their values (72, 182, and 192) range of ACL values. are more similar to the range of angiosperm values than to 166 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179

Fig. 2. Box plots divided by plant type for (A) average chain length (ACL), (B) carbon preference index (CPI), (C) relative percentage of the most dominant chain length of each n-alkane distribution (Dominance), (D) C31/(C29 +C31), (E) C31/(C27 +C31), (F) C23/(C23 +C29), (G) C33/(C29 +C33), (H) C35/(C29 +C35). (A–C) Show results for all moss data; (D–H) Report data for Sphagnum mosses only. Sample size of each type in parentheses. Each box represents range of middle 50% of group values, where center line is group mean. Whiskers are outside 25% each, and points are outliers. R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 167

et al., 2000). Comparisons were made first between woody plants and graminoids (a group which includes grasses as well as grass-like sedges and rushes: Poaceae, Cyperaceae, and Juncaceae families, respectively) and then secondly between woody plants and Sphagnum mosses. The interpretation of n-alkane chain-length ratios in terms of plant community rests on a foundational assumption of equivalent n-alkane production by different plant groups. Within the angiosperms, using observations with published n-alkane masses, we found that woody angiosperms (n = 85) and graminoids (n = 138) produce statistically equivalent amounts of n-alkanes in their leaf tissues (Student’s two- tailed t-test p = 0.0141; Fig. A-2). n-Alkane production appears to be much higher in angiosperms than gymnosperms (Fig. 4), and thus we included only woody angiosperms in the comparisons, but results for conifers (woody gymnosperms) were similar. First, we plotted groups on ternary diagrams, using either C27,C29, and C31 or C23,C27, and C29 as vertices (Fig. 5). Second, we performed discriminant Fig. 3. Box plots of average chain length (ACL) for temperate analyses for each set of plant groups, using the percentage (40°–60° latitude) woody plants: (A) Temperate deciduous and of two or more chain lengths as the determinant variables evergreen woody gymnosperms, (B) Temperate deciduous and (Tables 2 and 3, A-1). Average group chain-length profiles evergreen woody angiosperms. Note that 97 of 101 evergreen woody angiosperms are from the Ericaceae family. Sample size of are also presented (Fig. 6). each group in parentheses. 3.4.1. Woody plants and graminoids: C27,C29, and C31 gymnosperms. Fig. 4 also demonstrates that gymnosperms In order to approximate a geographically realistic have a significantly lower average CPI value than angio- assessment of plant groups, the data were divided into sperms: 5.20 (n = 296, std. dev. = 6.33) and 11.76 broad geographic distributions. One such geographic divi- (n = 1380, std. dev. = 12.25), respectively (Student’s t-test, sion includes all temperate zone plants, 40–60° latitude, p < 0.00005). from both hemispheres. Because of the large number of measurements made on temperate species in the Ericaceae 3.4. n-Alkane chain lengths and ratios as plant proxies (n = 128, of 188 temperate woody angiosperms), largely due to the work of Salasoo (1981, 1983, 1987a, 1987b, In addition to plotting ratios by plant group (Fig. 2D– 1987c), this family was separated from all other woody H), two other approaches were employed to examine the angiosperms. Fig. 5A shows the ternary diagram of C27, applicability of previously posited generalities that C31 pre- C29, and C31 abundances for temperate woody angio- dominates in grasses and C27 and C29 predominate in woo- sperms, excluding Ericaceae, alongside C3 and C4 grami- dy plants, i.e. trees and shrubs (Meyers and Ishiwatari, noids and the high degree of overlap between all three 1993; Meyers, 2003), and similarly that C23 and C25 pre- groups. Sub-Saharan Africa has the most n-alkane mea- dominate in Sphagnum mosses (Corrigan et al., 1973; Nott surements of all tropical and sub-tropical areas, and so

Fig. 4. Log–log plot of CPI values by n-alkane amounts for both angiosperms (ﬁlled diamonds, n = 196) and gymnosperms (open diamonds, n = 109). 168 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179

Fig. 5. Ternary diagrams of n-alkane chain-length abundances by plant group. All temperate woody angiosperms exclude the family

Ericaceae, except in 5G, where they are plotted separately. (A) Temperate woody angiosperms and C3 and C4 graminoids by C27,C29, and C31. (B) African woody angiosperms and C4 graminoids by C27,C29, and C31. (C) Temperate woody angiosperms and C3 and C4 graminoids by C27,C29, and C33. (D) African woody angiosperms and C4 graminoids by C27,C29, and C33. (E) Temperate woody angiosperms and C3 and C4 graminoids by C27,C29, and C35. (F) African woody angiosperms and C4 graminoids by C27,C29, and C35. (G) Temperate woody angiosperms, Ericaceae, and Sphagnum by C23,C27, and C29 n-alkanes. R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 169

Table 2

Results of discriminant analyses for graminoids and woody angiosperms using C29 and C31. True plants are actual group assignments, while determined (Det.) plants are statistically assigned. The integer indicates the number of observations for each category, and the number in parentheses indicates the percentage for each category from the row total. The percentage in parentheses for the grand total indicates the percentage of correctly assigned plants for both groups. Det. Graminoids Det. Woody Ang. Total

(A) All plants, classiﬁed using C29 and C31 as discriminants True Graminoids 200 (67%) 98 (33%) 298 True Woody Ang. 273 (46%) 318 (54%) 591 Total 473 (53%) 416 (47%) 889 (58%) (B) Temperate plants (40°–60° latitude) True Graminoids 93 (51%) 91 (49%) 184 True Woody Ang. 100 (53%) 88 (47%) 188 Total 193 (52%) 179 (48%) 372 (49%) (C) Sub-Saharan African plants True Graminoids 60 (94%) 4 (6%) 64 True Woody Ang. 33 (38%) 54 (62%) 87 Total 93 (62%) 58 (38%) 151 (75%)

the same comparisons between plant groups were also made tion: out of 298 total “true” graminoids, 200 (67%) were using African data only (Fig. 5B). Fig. 5C–F show addi- correctly determined to be graminoids while 98 (33%) were tional ternary diagrams constructed for the same temperate statistically assigned to “woody angiosperms.” Similarly, of and African plant groups using chain lengths C27,C29, and 591 total woody angiosperms, 318 (54%) were correctly C33 as well as C27,C29, and C35. These plots (Fig. 5C–F) identified and 273 (46%) were incorrectly identified as show less overlap between woody angiosperms and grami- “graminoids” by the discriminant analysis. The same anal- noids than the C31-based plots (Fig. 5A and B) based ysis was done for temperate plants from 40° to 60° latitude mostly on the absence of the longer C33 and C35 chain (Table 2B) and also for plants from sub-Saharan Africa lengths in many woody angiosperms. Nonetheless, there is (Table 2C). The discriminant chain lengths are capable of still a wide distribution of woody angiosperm values. determining sub-Saharan African graminoids with 94% To test the predictive power of the C29 and C31 chain accuracy but also incorrectly identify 38% of woody angio- lengths among plant groups, discriminant analyses were sperms as “graminoids”. In the real world, the true identi- performed (Table 2). Relative abundances of chain lengths ties are not known, and therefore the success of the C29 and C31 were used as discriminant variables to classify approach must be measured by the proportion of both plants as woody angiosperms and graminoids (both C3 and groups that are correctly identified. Thus, when both woo- C4). To do this, plants were treated as unknowns and statis- dy angiosperm and graminoid groups are considered to- tically assigned to either woody angiosperms or to grami- gether, the discriminant analyses are never better than noids based on the proximity of their abundances of C29 75% accurate for any of the geographic areas. The amount and C31 to the group means. Table 2A shows the results of variability in chain-length abundances is always higher of this analysis for all plants regardless of geographic loca- for woody angiosperms than for graminoids (Fig. 2C), and therefore the discriminant analysis is less accurate for Table 3 woody angiosperms in every case (Table 2). Discriminant Results of discriminant analyses for Sphagnum mosses and analyses utilizing C27 fared no better than those using C29 temperate woody angiosperms. True plants are actual group and C31 alone with the total correct assignments ranging assignments, while determined (Det.) plants are statistically from 53% (temperate) to 76% (sub-Saharan Africa) assigned. The integer indicates the number of observations for (Table A-1). Fig. 6A shows the average profiles for temper- each category, and the number in parentheses indicates the ate woody angiosperms (with Ericaceae excluded) as well as percentage for each category from the row total. The percentage temperate C3 and C4 graminoids, and Fig. 6B shows the in parentheses for the grand total indicates the percentage of average profiles for African woody angiosperms and C correctly assigned plants for both groups. 4 graminoids. C31/(C29 +C31) and C31/(C27 +C31) ratio val- All plants, classified using C23 and C29 as discriminants ues for all plants, temperate plants, and plants from Africa Det. Sphagnum Det. Woody Ang. Total as well as North America are provided in Tables A-2 and True Sphagnum 39 (83%) 8 (17%) 47 A-3. True Woody Ang. 6 (3%) 182 (97%) 188 To explore whether the apparent separation between Total 45 (19%) 190 (81%) 235 (94%) African woody angiosperms and African C4 graminoids 170 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179

Fig. 6. Average n-alkane distributions. (A) Temperate woody angiosperms (Ericaceae excluded), temperate C3 graminoids, and temperate C4 graminoids. (B) African woody angiosperms and C4 graminoids. (C) Temperate Sphagnum mosses, woody angiosperms (Ericaceae excluded), and Ericaceae species. Numbers in parentheses are sample sizes. Error bars are ±1 standard deviation.

(Figs. 5B, D, F and 6) was related to growing environment (Fig. A-3B and C, respectively) rainforest woody angio- in Africa, woody angiosperms published in Vogts et al. sperms as a group have lower amounts of these longer chain (2009) were separated into either rainforest or savanna bio- lengths than any of the savanna plants. Discriminant anal- mes (following the categories listed by the authors) and ysis was also performed for the rainforest versus savanna plotted on ternary diagrams alongside savanna forbs and woody angiosperms only, using C29 and C31. The analysis C4 graminoids from eastern or southern Africa (Fig. A-3). correctly identiﬁed 22 of 24 rainforest woody plants and Fig. A-3A shows that for C31, there is still a large amount 37 of 45 savanna woody plants, for a total accuracy of of overlap among all groups, including both rainforest 86% (Table A-4). Including C33 and C35 in the discriminant and savanna plants, but for chain lengths C33 and C35 analysis also produced 86% total accuracy. R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 171

3.4.2. Sphagnum mosses and woody plants: C23,C27, and C29 the relative chain-length proportions remain fairly steady, Fig. 2F demonstrates the differentiation between Sphag- and Tipple et al. (2013) show that both ACL and isotopic num and other plant groups based on a C23/(C23 +C29) ra- values reflect the time of leaf formation and remain steady tio. The ternary diagram in Fig. 5G also visually through the rest of the growing season. Sampling of imma- demonstrates the ability of C23,C27, and C29 chain-length ture leaves may also explain the variation observed by abundances to separate Sphagnum mosses from woody Strańsky´ et al. (1967). Lastly, Avato et al. (1984) also found angiosperms. Because the Ericaceae plant family is an that once leaves mature, n-alkane quantities remain rela- important component of the bog wetland ecosystems where tively constant. The results presented here suggest that for Sphagnum species predominate, we have plotted woody fully developed leaves, there is no significant difference in angiosperms as two groups, one excluding Ericaceae, simi- relative n-alkane distributions either around a tree or within lar to the comparison with graminoids, and the other the a growing season. It should be noted that Maffei et al. Ericaceae family itself (Fig. 5G). The two groups of woody (1993) found that n-alkanes in the evergreen Rosmarinus angiosperms overlap one another, and the Sphagnum officinalis correlated with temperature when sampled across mosses are distinguished from woody angiosperms almost summer and winter, suggesting that for evergreen leaves n- entirely by their high proportion of C23. Table 3 shows alkane composition shifts to longer chain lengths in re- the results of a discriminant analysis using relative abun- sponse to lower winter temperatures in overwintering dances of C23 and C29 to differentiate temperate woody leaves. However, for the outdoor trees measured in this angiosperms from Sphagnum. Because Ericaceae and other study across a summer growing season, n-alkane distribu- woody angiosperms were so similar in their distributions of tion does not significantly change from at least the point C23,C27, and C29, the two groups were not separated from of leaf maturation to leaf abscission in the autumn one another in the discriminant analysis. Results of discrim- (Fig. 1B). Although these data come from a limited number inant analyses using all three chain-length abundances (C23, of species from a non-natural (although outdoor) environ- C27, and C29) and of only C23 and C27 are not shown be- ment, the strong correlation between sun and shade as well cause they are exactly the same as for C23 and C29. The as summer and fall n-alkane distributions supports chain lengths can be used to separate the two plant groups comparisons of the distributions of modern plants with from one another with overall 94% accuracy (Table 3). fossil n-alkanes of unknown canopy position and season n-Alkane ratios were also calculated for woody plants and of abscission, and in this sense makes them well suited for Sphagnum mosses, with C23 and C25 used to represent use as paleoecological proxies. Sphagnum and C27 and C29 to represent woody plants. Ratios using C23 (Table A-4) provided a better proxy for separating 4.2. n-Alkane distributions across plant functional types mosses from woody plants than did ratios using C25 (Tables A-5 and A-6). Sphagnum mosses had C23/(C23 +C27) and Average chain length is a broad-brush parameter that C23/(C23 +C29) ratios of 0.773 and 0.728, respectively, while does little to differentiate large-scale plant groups for any those for woody angiosperms were 0.089 and 0.071. Data for group except Sphagnum mosses (Fig. 2A). Some species ex- Sphagnum came only from temperate regions and so were hibit a high degree of genetic control over their n-alkane compared only with temperate woody plants. Fig. 6C shows distributions, such that different plants growing in different the average profiles for Sphagnum mosses, woody angio- locations can have relatively similar patterns of n-alkane sperms excluding Ericaceae, and Ericaceae. distribution, e.g. the almost exclusive production of C27 by Fagus sylvatica leaves (Gu¨lz et al., 1989; Lockheart 4. DISCUSSION et al., 1997; Tu et al., 2007; Sachse et al., 2009; this study). However, other species have been shown to have variable n- 4.1. n-Alkane distributions within trees alkane distributions across different environments, limiting the usefulness of n-alkane profiles in chemotaxonomy n-Alkane distributions for temperate trees from the (Dodd and Poveda, 2003; Vogts et al., 2009). Because the Chicago Botanic Garden do not appear to significantly vary leaf wax is a plant’s first barrier to the external environ- with either canopy position or with sampling date, for both ment, it is reasonable to hypothesize that the composition angiosperms and gymnosperms (Fig. 1). The stability of of that wax, including the n-alkanes, would be controlled n-alkane distributions over a growing season shown here to some degree by environmental adaptation and therefore is in apparent contrast to the findings of other researchers, plastic in response to external conditions (Shepherd and who describe changes in n-alkane amounts as leaves mature Griffiths, 2006). This plasticity would contribute to the ob- across a growing season (Piasentier et al., 2000). However, served variation and lack of differentiation among plant Piasentier et al. (2000) show that the greatest changes in rel- groups. ative n-alkane amounts occur during the growth and expan- Regarding CPI, the vast majority (but not all) of mea- sion of the young leaf, and they note that the n-alkane sured modern plants have values greater than 1 (Fig. 2B), patterns of the mature and senescing leaves (which corre- meaning that odd chain lengths are more abundant than spond to the leaves sampled in this study) remain markedly even chain lengths. Perhaps not surprisingly, a reasonable similar, in agreement with the findings presented here. Sim- cut-off for determining a relatively unmodified terrestrial ilarly, although Sachse et al. (2009) report variability in leaf plant source for sediment n-alkanes appears to be a CPI n-alkane amounts across multiple dates in a single growing of 1 or greater (Douglas and Eglinton, 1966), and a more season, suggesting ablation and turnover of surface waxes, rigorous threshold might be 2 or greater, as the majority 172 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 of modern plants (81% of 1723 measurements) fall above 2. (e.g. Nichols et al., 2006; Smith et al., 2007; Diefendorf However, in light of the tremendous diversity of CPI values et al., 2010). However, the demonstrated difference in n-al- in modern plants, it seems prudent to use CPI as an indica- kane production by angiosperms and gymnosperms re- tor of terrestrial plant origin in a more binary (either the n- quires careful interpretation of n-alkanes from alkanes are faithful to the original plant source or they are sedimentary archives (Fig. 4; Diefendorf et al., 2011). not) sense rather than interpreting along a spectrum of val- Although n-alkane production in Sphagnum and angio- ues. It seems especially difficult to apply sedimentary CPI sperms appears to be more similar, additional data on abso- values in any quantitative way towards determining the de- lute amounts of n-alkane chain lengths in modern plants gree of n-alkane degradation as in Buggle et al. (2010) and (lg n-alkane/g dry leaf) in woody, herbaceous, and Sphag- Zech et al. (2013), although they may still qualitatively indi- num-dominated ecosystems are needed. cate thermal maturity in oils and petroleum source rocks (Bray and Evans, 1961). The use of CPI in combination 4.3. n-Alkane chain lengths as grass and woody plant proxies with n-alkane ratios to estimate the n-alkane contributions of graminoids and trees (discussed in Section 4.3), e.g. fol- The initial studies that are often cited in support of the lowing Zech et al. (2009), should be viewed with particular C31/C29 grass/woody plant comparison make no mention caution or even be disregarded due to the extreme variabil- of grasses separate from other terrestrial plants, but rather ity and overlap in these parameters in modern plants focus on terrestrial plants as a group in comparison with (Fig. 2). aquatic plants or peat (Cranwell, 1973, 1984; Cranwell Lastly, both deciduous and evergreen gymnosperms et al., 1987). The compilation of published data and new produce, on average, much smaller quantities of n-alkanes analyses presented here demonstrate that both graminoids per unit leaf mass than do angiosperms. It was observed and woody angiosperms produce C29 and C31 in abundance previously that in some individuals of the gymnosperms relative to other n-alkane chain lengths (Fig. 6A), and the ra- Pinaceae and Cupressaceae, n-alkanes constituted only tios of these two compounds are highly variable and over- 1% of leaf wax (Herbin and Robins, 1968; Chikaraishi lapping between these groups (Fig. 2D). Therefore, C29 et al., 2004), while leaf waxes of some angiosperm species and C31 cannot serve as generalized proxies for separating were over 90% n-alkanes (Herbin and Robins, 1969; Dove, grasses and woody plants. The ternary diagrams demon- 1992). Diefendorf et al. (2011) also found in their measure- strate visually the large amount of variation within the ments of leaf n-alkanes from Pennsylvania, Wyoming, and groups, especially in woody angiosperms, as well as the Panama that angiosperms consistently produced more n-al- overlap of graminoids and woody angiosperms (Fig. 5A kanes than gymnosperms. Here we expand on these find- and B). Discriminant analyses using C29 and C31, with or ings by demonstrating this difference across 274 species without C27, are at worst little better than random guessing and a range of plant families and locations (from Africa, at correctly identifying graminoids and woody angiosperms Europe, Australia, and South America) (Fig. 4). This find- by their chain-length abundances (49% for temperate plants, ing in modern plants fits with the observation by Lockheart 58% correct globally) and are at best 75% accurate for trop- et al. (2000) and Otto et al. (2005) that gymnosperm fossils ical regions (Table 2), mostly because of the large amount of had smaller amounts of n-alkanes than did angiosperm fos- variation in woody angiosperm n-alkane abundances. sils from the same localities. For paleoecological applica- A possible exception is the regional grouping of Sub- tions, this difference in n-alkane production means that Saharan graminoids, which appear to be distinguishable even for an ecosystem with a minority of angiosperms based on C29 and C31 relative abundances (Table 2C, and a majority of gymnosperms, angiosperms would likely Fig. 5B). However, this apparent association between chain be the dominant source of in situ sediment n-alkanes in the lengths and plant groups may instead reflect the influence of local soil organic matter. For example, if we consider an environment concealed by inadvertent sampling bias. Most ecosystem that is 90% gymnosperm and only 10% angio- of the sub-Saharan graminoids are C4 plants that are sperm and we use the average amounts of n-alkanes re- adapted to hot, dry environments (Rommerskirchen ported for each group here (510 lg n-alkane/g dry leaf in et al., 2006). By comparison, many of the woody angio- angiosperms and 16 lg in gymnosperms, excluding the out- sperm observations are drawn from a greater diversity of lying Podocarpus measurement, Fig. 4), the mixed sedimen- environments (hot and dry as well as cooler and wetter, tary n-alkanes would be roughly 22% gymnosperm and 78% e.g. Vogts et al., 2009). It seems plausible that the greater angiosperm in origin. If we consider an ecosystem that is an predominance of longer chain lengths (C31,C33, and C35) even mix of angiosperms and gymnosperms, angiosperms in graminoids than in woody angiosperms in Africa would produce 97% of the total n-alkane signal. The differ- (Figs. 5B, D, F and 6) could be a more direct reflection ence in n-alkane production between angiosperms and gym- of local climatic influences such as aridity than of the plant nosperms observed here for total n-alkanes is of a similar groups per se (Rommerskirchen et al., 2006). To evaluate magnitude to that reported by Diefendorf et al. (2011). this possibility, we examined the n-alkanes from woody Sphagnum mosses appear to produce n-alkanes in their tis- angiosperms sampled from both African rainforest and sa- sues in greater quantities than most gymnosperms, but still vanna biomes in the study by Vogts et al. (2009). Savanna less than those of angiosperm leaves (Pancost et al., 2002). woody angiosperms have relative abundances of the longer Inferences of plant cover from distributions and stable iso- chain lengths (C31,C33, and C35 relative to C29 and C27) tope ratios of n-alkanes often implicitly assume that plants similar to the C4 graminoids and forbs (Fig. A-3). In contribute n-alkanes in direct proportion to their biomass contrast, rainforest woody plants group separately from R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 173 the savanna woody angiosperms, graminoids, and forbs, or woody angiosperm. Most woody angiosperms also have suggesting that climate could be the controlling factor more C29 than C27, which supports the use of comparisons rather than woody versus herbaceous life habit (Fig. A-3). between C23 and C29 as a proxy for inputs of Sphagnum Furthermore, when considering only woody angiosperms moss and woody plants (Nichols et al., 2006). However, sub- (i.e. controlling for plant type), a discriminant analysis merged and floating aquatic plants have also been demon- using C29 and C31 is able to correctly identify plants from strated to contain a relative abundance of C23 and C25 the two separate biomes (rainforest and savanna) with (Ficken et al., 2000; Mu¨gler et al., 2008), and therefore, as 86% accuracy (Table A-4). Interestingly, C35 relative to in any consideration of ecological proxy indicators, the envi- C29 and C27 appears to largely separate graminoids and for- ronmental context—e.g. of a Sphagnum-dominated bog or a bs from savanna woody angiosperms (Fig. A-3C), but fur- lacustrine system lacking in Sphagnum but with aquatic ther investigation is required to determine whether this macrophytes present—is crucial. Nonetheless, the data pre- pattern reflects plant type or differences in microclimates sented here demonstrate that for bog ecosystems, this is a within the savanna (e.g. woody species sampled from ripar- useful proxy for estimating the proportion of n-alkanes from ian environments). African graminoids may occupy a more Sphagnum mosses or from woody plants (Nott et al., 2000; restricted range of chain-length distributions (Fig. 5), simi- Nichols et al., 2010). lar to savanna forbs and woody plants and in contrast to rainforest woody plants (Fig. A-3), because their environ- 4.5. Implications for paleoecology mental range is more restricted (i.e. to warmer, more arid environments) than woody plants, which can be found in From a paleoecological perspective, the consistency of n- some capacity in almost any environment. alkane chain-length patterns within mature leaves over time Other studies have also suggested correlations between and canopy position validates the comparison of n-alkanes n-alkane chain lengths and environment (Kawamura from modern plants with sediment or fossil n-alkanes that et al., 2003; Rommerskirchen et al., 2003; Sachse et al., derive from leaves of unknown provenance. However, the 2006). For example, Castan˜eda et al. (2009) examined sev- tremendous range of CPI values in living plants makes eral molecular proxies from sediment cores in Lake Malawi any application for sediment CPI values beyond a rough and showed that n-alkane chain length was correlated with qualitative assessment of source fidelity particularly prob- temperature reconstructions and did not correlate with lematic. Genetic controls clearly exert some degree of influ- reconstructions of grass and woody plant dominance based ence over the production of different n-alkane chain lengths, on lignin phenolic compounds. Vogts et al. (2012) found as demonstrated by chemotaxonomic studies (Vioque et al., that ACL values from marine sediments off the coast of 1994; Maffei et al., 2004; Medina et al., 2006; Bingham et al., western Africa correlated with continental aridity and that 2010) and biosynthetic studies (Jetter et al., 2006; Kunst ACL-based reconstructions of C4 plant abundances et al., 2006). Similarly, the relative abundances of n-alkanes matched d13C-based reconstructions. Studies of n-alkane appear to be relatively tightly controlled within a single distributions in living plants (Dodd and Poveda, 2003; plant, across both canopy and growing season (Fig. 1). Be- Sachse et al., 2006), as well as studies of aeolian dust and yond the individual plant, it remains unclear why one plant marine sediments from the Atlantic and Pacific (Rinna group per se (e.g. graminoids or woody plants growing in the et al., 1999), have found correlations between n-alkane same environment) would favor production of one chain chain lengths and latitude or altitude, suggesting instead length over another. The variation in chain-length abun- that there may be a relationship between longer chain dances within most groups is large, even when accounting lengths and warmer, drier, or possibly more irradiated for factors such as region and photosynthetic pathway environments. (Figs. 2 and 5). This makes it inadvisable to use n-alkane chain-length abundances as chemotaxonomic indicators 4.4. n-Alkane chain lengths as a Sphagnum proxy for broad plant functional groups, with the exception of Sphagnum mosses from peat bog ecosystems. Based on the In contrast to the comparison between graminoids and modern plant data, ratios of chain lengths C27 or longer, woody plants, ratios of the shorter C23 n-alkane to either such as C31/(C29 +C31), do not appear to be a robust proxy C27 or C29 appear to readily distinguish modern Sphagnum for separating grasses from woody plants. By contrast, C23 mosses from woody angiosperms. Indeed, it seems reason- appears to generate a generally reliable proxy for Sphagnum able to expand this conclusion to encompass all non-Sphag- mosses when compared against C27 or C29 as indicators of num terrestrial plants, including grasses, forbs, and input from other terrestrial plants, which is consistent with gymnosperms, because of the lack of distinction among all the findings of others (Pancost et al., 2002; Nichols et al., of the other plant groups from one another (Fig. 2F). Sphag- 2006, 2010). With further investigation into the absolute num and woody angiosperms are largely distinguished from quantities of n-alkanes produced per plant biomass and cov- one another by the relative abundance of C23 (Fig. 5G): er, it appears that these n-alkane chain lengths could poten- Sphagnum samples in the ternary diagrams have abundances tially be a semi-quantifiable proxy for plant group cover in greater than 25% C23, compared to C27 and C29, while most Sphagnum wetlands (Nichols et al., 2010) or for Sphag- woody angiosperms (including the Ericaceae) have less than num-sourced organic matter in river and marine sediments 10% C23. The results of the discriminant analyses utilizing (Vonk et al., 2008; Vonk and Gustafsson, 2009). C23 and C29 as well as C23 and C27 both showed 94% overall Environmental factors may play a role in the plasticity accuracy in correctly assigning groups to either Sphagnum of n-alkane distributions and explain some of the variation 174 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 in large-scale plant groups. It is possible that climate can canopy and growing season, and these findings validate the independently drive both plant community composition comparison between modern plant n-alkanes and n-alkanes and n-alkane chain-length distribution, and that the two from fossil soils and sediments derived from leaves of un- phenomena could be correlated without a direct causal rela- known provenance. Furthermore, consistent with other re- tionship. Most sedimentary n-alkane archive studies report cent findings (Diefendorf et al., 2011 and references concurrent shifts in local climate and plant communities, therein), the evidence here shows that angiosperms produce e.g. the large climate shifts associated with glacial/intergla- orders of magnitude more n-alkanes in their leaves than do cial cycles (Brincat et al., 2000; Zhang et al., 2006; Zech gymnosperms: typically 100s to 1000s of lg n-alkanes per g et al., 2009) and with the Paleocene-Eocene Thermal Max- dry leaf matter compared to 10s or less, respectively. We imum (Smith et al., 2007). It is difficult to disentangle also demonstrate that n-alkane chain lengths can serve as whether the observed changes in n-alkane distributions useful chemotaxonomic proxies for the identification of through time are due to the direct influence of climate, Sphagnum mosses from other plant groups, but are unable based on the plastic response of n-alkanes to temperature to distinguish graminoids from woody plants, and that or aridity, or to its indirect influence via climate-driven modern plants exhibit a tremendous range of n-alkane ratio shifts in the local plant community. However, given the and CPI values. Previous studies have suggested that n-al- range of variation within plant groups, another possibility kane chain-length distributions may be influenced by envi- remains that shifts in sediment n-alkane distributions may ronment, possibly in addition to genetic controls. Thus, it is reflect species turnover in plant communities that is de-cou- possible that coherent patterns of n-alkanes in sediment re- pled both from plant types and from a direct climate driver cords are not artifactual but rather may directly result from (e.g. due to succession, competition, etc.) Considering the climate forcings, meaning that there is present potential for low amounts of n-alkanes synthesized in gymnosperm n-alkane chain-length distributions to serve as a paleocli- leaves, the gymnosperm contribution to sediment n-alkanes mate proxy. However, this possibility must first be directly may only rarely be significant (e.g. where pollen or leaf fos- tested in modern plants and sediments. Examination into sil evidence indicates a preponderance of gymnosperms in the biosynthetic controls on n-alkane production and the the local plant community). Future research would benefit manifestation of n-alkane response to climatic influences from a determination of the environmental controls on bio- will further elucidate the ecological interpretations possible synthesis of n-alkanes and the extent to which climate plays with these valuable plant biomarkers. a role in determining n-alkane chain lengths in modern plants and sediments. ACKNOWLEDGMENTS

5. CONCLUSIONS We thank the research and horticultural staff and especially Nyree Zerega and Celeste VanderMey at the Chicago Botanic Gar- den for their facilitation in sample collection. We are grateful to n-Alkanes have been acknowledged for decades as useful Sarah Feakins for sharing data, and to Ellen Currano, Anna Hen- biomarkers for terrestrial plants, and dozens of studies have derson, Phil Meyers, Scott Wing, and three anonymous reviewers examined the distributions of n-alkane chain lengths across for comments. Funding was provided by EPA Science to Achieve hundreds of different plant species. Until now, however, Results (STAR) Graduate Fellowship (to R.T.B.), research grants most of these studies have focused on single taxonomic from the Plant Biology and Conservation graduate program at groups or geographic areas, and broad paleoecological Northwestern University and the Chicago Botanic Garden (to interpretations have been based on a relatively narrow R.T.B.), the Initiative for Sustainability and Energy at Northwest- and incomplete set of modern data. Here, we find that in ern (ISEN) (to R.T.B. and F.A.M.) and the Australian Research temperate trees n-alkane distributions are consistent across Council FT110100793 (to F.A.M.).

Table A-1

Results of discriminant analyses for graminoids and woody angiosperms using C27,C29, and C31. True plants are actual group assignments, while determined (Det.) plants are statistically assigned. The integer indicates the number of observations for each category, and the number in parentheses indicates the percentage for each category from the row total. The percentage in parentheses for the grand total indicates the percentage of correctly assigned plants for both groups. Det. Graminoids Det. Woody Ang. Total

(A) All plants, classiﬁed using C27,C29, and C31 as discriminants True Graminoids 196 (66%) 102 (34%) 298 True Woody Ang. 287 (49%) 304 (51%) 591 Total 483 (54%) 406 (46%) 889 (56%) (B) Temperate plants (40°-60° latitude) True Graminoids 115 (63%) 69 (38%) 184 True Woody Ang. 104 (55%) 84 (45%) 188 Total 219 (59%) 153 (41%) 372 (53%) (C) Sub-Saharan African plants True Graminoids 55 (86%) 9 (14%) 64 True Woody Ang. 27 (31%) 60 (69%) 87 Total 82 (54%) 69 (46%) 151 (76%) R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 175

Table A-2 Table A-4

Average C31/(C29 +C31) ratios for plant types, including sample Results of discriminant analyses for African rainforest versus size (n) and standard deviation (SD). savanna woody angiosperms from Vogts et al. (2009). True plants C /(C +C ) n Average SD are actual group assignments, while determined (Det.) plants are 31 29 31 statistically assigned. The integer indicates the number of obser- All plants vations for each category, and the number in parentheses indicates Conifers 219 0.642 0.303 the percentage for each category from the row total. The Woody Angiosperms 585 0.479 0.289 percentage in parentheses for the grand total indicates the Graminoids 298 0.610 0.194 percentage of correctly assigned plants for both groups. C3 Graminoids 150 0.559 0.206 C Graminoids 144 0.660 0.167 African woody angiosperms, classiﬁed using C29 and C31 as 4 discriminants Temperate (40–55° latitude) Det. Rainforest Det. Savanna Total Conifers 170 0.643 0.314 Woody Angiosperms 188 0.484 0.257 True Rainforest 22 (92%) 2 (8%) 24 Woody Angiosperms (excl. Ericaceae) 60 0.359 0.283 True Savanna 8 (18%) 37 (82%) 45 Ericaceae 128 0.542 0.223 Total 30 (43%) 39 (57%) 69 (86%) Graminoids 184 0.555 0.196

C3 Graminoids 125 0.553 0.210 C4 Graminoids 57 0.556 0.164 Table A-5 Average C /(C +C ), C /(C +C ), and C /(C +C ) Africa 23 23 27 23 23 29 23 23 31 ratios for woody angiosperms and Sphagnum mosses, including Woody Angiosperms 87 0.399 0.270 sample size (n) and standard deviation (SD). Graminoids 64 0.731 0.113

C3 Graminoids 1 0.645 – n Average SD C Graminoids 63 0.732 0.114 4 C23/(C23 +C27) North America Temperate (40–55° latitude) Woody Angiosperms 158 0.418 0.292 Woody Angiosperms 188 0.089 0.157 Woody Angiosperms (excl. Ericaceae) 83 0.291 0.310 Woody Angiosperms (excl. 60 0.132 0.183 Ericaceae 75 0.558 0.191 Ericaceae) Graminoids 46 0.601 0.177 Ericaceae 128 0.069 0.139 C3 Graminoids 18 0.664 0.151 Sphagnum 47 0.773 0.120 C4 Graminoids 27 0.551 0.179 C23/(C23 +C29) Temperate (40–55° latitude) Table A-3 Woody Angiosperms 188 0.071 0.166 Average C31/(C27 +C31) ratios for plant types, including sample Woody Angiosperms (excl. 60 0.099 0.153 size (n) and standard deviation (SD). Ericaceae) Ericaceae 128 0.057 0.170 C31/(C27 +C31) n Average SD Sphagnum 47 0.728 0.171 All plants Conifers 219 0.689 0.295 C23/(C23 +C31) Woody Angiosperms 581 0.685 0.341 Temperate (40–55° latitude) Graminoids 298 0.770 0.197 Woody Angiosperms 185 0.112 0.234

C3 Graminoids 150 0.790 0.205 Woody Angiosperms (excl. 57 0.180 0.240 Ericaceae) C4 Graminoids 144 0.749 0.186 Ericaceae 128 0.081 0.225 Temperate (40–55° latitude) Sphagnum 47 0.608 0.229 Conifers 171 0.684 0.310 Woody Angiosperms 188 0.693 0.319 Woody Angiosperms (excl. Ericaceae) 60 0.516 0.342 Table A-6 Ericaceae 128 0.776 0.272 Average C25/(C25 +C27), C25/(C25 +C29), and C25/(C25 +C31) Graminoids 184 0.757 0.217 ratios for woody angiosperms and Sphagnum mosses, including sample size (n) and standard deviation (SD). C3 Graminoids 125 0.794 0.208 C4 Graminoids 57 0.680 0.211 n Average SD

Africa C25/(C25 +C27) Woody Angiosperms 87 0.579 0.338 Temperate (40–55° latitude) Graminoids 64 0.792 0.134 Woody Angiosperms 188 0.210 0.173

C3 Graminoids 1 0.826 – Woody Angiosperms (excl. 60 0.248 0.199 C4 Graminoids 63 0.791 0.135 Ericaceae) Ericaceae 128 0.193 0.158 North America Sphagnum 47 0.747 0.089 Woody Angiosperms 154 0.646 0.387 Woody Angiosperms (excl. Ericaceae) 79 0.498 0.435 C25/(C25 +C29) Ericaceae 75 0.803 0.250 Temperate (40–55° latitude) Graminoids 46 0.748 0.235 Woody Angiosperms 188 0.139 0.203 C3 Graminoids 18 0.896 0.125 (continued on next page) C4 Graminoids 27 0.641 0.237 176 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179

Table A-6 (continued) APPENDIX A. Woody Angiosperms (excl. 59 0.341 0.332 Ericaceae) Ericaceae 128 0.105 0.183 Sphagnum 47 0.695 0.151 APPENDIX B. SUPPLEMENTARY DATA C /(C +C ) 25 25 31 Supplementary data associated with this article can be Temperate (40–55° latitude) Woody Angiosperms 187 0.197 0.293 found, in the online version, at http://dx.doi.org/10.1016/ Woody Angiosperms (excl. 69 0.350 0.326 j.gca.2013.04.016. Ericaceae) Ericaceae 128 0.132 0.249 Sphagnum 47 0.572 0.206

Fig. A-1. ACL calculated using only odd chain length n-alkanes (Odd ACL) compared with ACL calculated using both odd and even chain lengths (Full ACL). Dashed line is 1:1 line. Solid line is linear regression line.

Fig. A-2. Log–log plot of CPI by n-alkane amounts across angiosperm habit groups for which data was available. R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 177

REFERENCES

Avato P., Bianchi G. and Mariani G. (1984) Epicuticular waxes of Sorghum and some compositional changes with plant age. Phytochemistry 23, 2843–2846. Bingham E. M., McClymont E. L., Va¨liranta M., Mauquoy D., Roberts Z., Chambers F. M., Pancost R. D. and Evershed R. P. (2010) Conservative composition of n-alkane biomarkers in Sphagnum species: implications for palaeoclimate reconstruction in ombrotrophic peat bogs. Org. Geochem. 41, 214–220. Bray E. E. and Evans E. D. (1961) Distribution of n-paraffins as a clue to recognition of source beds. Geochim. Cosmochim. Acta 22, 2–15. Brincat D., Yamada K., Ishiwatari R., Uemura H. and Naraoka H. (2000) Molecular isotopic stratigraphy of long-chain n- alkanes in Lake Baikal Holocene and glacial age sediments. Org. Geochem. 31, 287–294. Brooks J. R., Flanagan L. B., Buchmann N. and Ehleringer J. R. (1997) Carbon isotope composition of boreal plants: functional grouping of life forms. Oecologia 110, 301–311. Buggle B., Wiesenberg G. L. B. and Glaser B. (2010) Is there a possibility to correct fossil n-alkane data for postsedimentary alteration effects? Appl. Geochem. 25, 947–957. Castan˜eda I. S. and Schouten S. (2011) A review of molecular organic proxies for examining modern and ancient lacustrine environments. Quatern. Sci. Rev. 30, 2851–2891. Castan˜eda I. S., Werne J. P., Johnson T. C. and Filley T. R. (2009) Late Quaternary vegetation history of southeast Africa: the molecular isotopic record from Lake Malawi. Palaeogeogr. Palaeoclimatol. Palaeoecol. 275, 100–112. Chibnall A. C., Piper S. H., Pollard A., Williams E. F. and Sahai P. N. (1934) The constitution of the primary alcohols, fatty acids and paraffins present in plant and insect waxes. Biochem. J. 28, 2189–2208. Chikaraishi Y., Naraoka H. and Poulson S. R. (2004) Carbon and hydrogen isotopic fractionation during lipid biosynthesis in a higher plant (Cryptomeria japonica). Phytochemistry 65, 323–330. Corrigan D., Kloos C., O’Connor C. S. and Timoney R. F. (1973) Alkanes from four species of Sphagnum moss. Phytochemistry 12, 213–214. Cranwell P. A. (1973) Chain-length distribution of n-alkanes from lake sediments in relation to post-glacial environmental change. Freshwat. Biol. 3, 259–265. Cranwell P. A. (1984) Lipid geochemistry of sediments from Upton Broad, a small productive lake. Org. Geochem. 7, 25–37. Cranwell P. A., Eglinton G. and Robinson N. (1987) Lipids of aquatic organisms as potential contributors to lacustrine sediments – II. Org. Geochem. 11, 513–527. Diefendorf A. F., Mueller K. E., Wing S. L., Koch P. L. and Freeman K. H. (2010) Global patterns in leaf 13C discrimina- tion 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. Dodd R. S. and Afzal-Rafii Z. (2000) Habitat-related adaptive properties of plant cuticular lipids. Evolution 54, 1438–1444. Fig. A-3. Ternary diagrams of African woody angiosperms from Dodd R. S. and Poveda M. M. (2003) Environmental gradients and Vogts et al. (2009), separated into rainforest and savanna biomes, population divergence contribute to variation in cuticular wax with African C4 graminoids and forbs. (A) C27,C29, and C31. (B) composition in Juniperus communis. Biochem. Syst. Ecol. 31, C27,C29, and C33.C.C27,C29, and C35. 1257–1270. 178 R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179

Douglas A. G. and Eglinton G. (1966) The distribution of alkanes. Kunst L., Jetter R. and Samuels A. L. (2006) Biosynthesis and In Comparative Phytochemistry (ed. T. Swain). Academic Press, transport of plant cuticular waxes. In Biology of the Plant London. Cuticle (eds. M. Reiderer and C. Mu¨ller). Blackwell Publishing, Dove H. (1992) Using the n-alkanes of plant cuticular wax to Oxford. estimate the species composition of herbage mixtures. Aust. J. Lockheart M. J., van Bergen P. F. and Evershed R. P. (1997) Agric. Res. 43, 1711–1724. Variations in the stable carbon isotope composition of individ- Dyson W. G. and Herbin G. A. (1968) Studies on plant cuticular ual lipids from the leaves of modern angiosperms: implications waxes – IV. Leaf wax alkanes as a taxonomic discriminant for for the study of higher land plant-derived sedimentary organic cypresses grown in Kenya. Phytochemistry 7, 1339–1344. matter. Org. Geochem. 26, 137–153. Eglinton G. and Hamilton R. J. (1963) The distribution of alkanes. Lockheart M. J., van Bergen P. F. and Evershed R. P. (2000) In Chemical Plant Taxonomy (ed. T. Swain). Academic Press, Chemotaxonomic classification of fossil leaves from the Mio- London. cene Clarkia lake deposit, Idaho, USA based on n-alkyl lipid Eglinton G. and Hamilton R. J. (1967) Leaf epicuticular waxes. distributions and principle component analyses. Org. Geochem. Science 156, 1322–1335. 31, 1223–1246. Eglinton G. and Logan G. A. (1991) Molecular preservation. Phil. Maffei M., Mucciarelli M. and Scannerini S. (1993) Environmental Trans. Roy. Soc. London B 333, 315–328. factors affecting the lipid metabolism in Rosmarinus officinalis Ficken K. J., Li B., Swain D. L. and Eglinton G. (2000) An n- L. Biochem. Syst. Ecol. 21, 765–784. alkane proxy for the sedimentary input of submerged/floating Maffei M., Badino S. and Bossi S. (2004) Chemotaxonomic freshwater aquatic macrophytes. Org. Geochem. 31, 745–749. significance of leaf wax n-alkanes in the Pinales (Coniferales). Greenwood D. R. (1991) The taphonomy of plant macrofossils. In J. Biol. Res. 1, 3–19. The Processes of Fossilization (ed. S. K. Donovan). Columbia Marzi R., Torkelson B. E. and Olson R. K. (1993) A revised University Press, New York. carbon preference index. Org. Geochem. 20, 1303–1306. Gu¨lz P.-G. and Mu¨ller E. (1992) Seasonal variation in the Medeiros P. M. and Simoneit B. R. T. (2007) Gas chromatography composition of epicuticular waxes of Quercus robur leaves. Z. coupled to mass spectrometry for analyses of organic com- Naturforsch. 47c, 800–806. pounds and biomarkers as tracers for geological, environmen- Gu¨lz P.-G., Mu¨ller E. and Prasad R. B. N. (1989) Organ-specific tal, and forensic research. J. Sep. Sci. 30, 1516–1536. composition of epicuticular waxes of beech (Fagus sylvatica L) Medina E., Aguiar G., Go´mez M., Aranda J., Medina J. D. and leaves and seeds. Z. Naturforsch. 44c, 731–734. Winter K. (2006) Taxonomic significance of the epicuticular Handley L., Pearson P. N., McMillan I. K. and Pancost R. D. wax composition in species of the genus Clusia from Panama. (2008) Large terrestrial and marine carbon and hydrogen Biochem. Syst. Ecol. 34, 319–326. isotope excursions in a new Paleocene/Eocene boundary section Meyers P. A. (2003) Applications of organic geochemistry to from Tanzania. Earth Planet Sci. Lett. 275, 17–25. paleolimnological reconstructions: a summary of examples Hanisch S., Ariztegui D. and Pu¨ttmann W. (2003) The biomarker from the Laurentian Great Lakes. Org. Geochem. 34, 261–289. record of Lake Albano, central Italy–implications for Holocene Meyers P. A. and Ishiwatari R. (1993) Lacustrine organic aquatic system response to environmental change. Org. Geo- geochemistry – an overview of indicators of organic matter chem. 34, 1223–1235. sources and diagenesis in lake sediments. Org. Geochem. 20, Herbin G. A. and Robins P. A. (1968) Studies on plant cuticular 867–900. waxes – III. The leaf wax alkanes and x-hydroxy acids of some Mu¨gler I., Sachse D., Werner M., Xu B., Wu G., Yao T. and members of the Cupressaceae and Pinaceae. Phytochemistry 7, Gleixner G. (2008) Effect of lake evaporation on dD values of 1325–1337. lacustrine n-alkanes: a comparison of Nam Co (Tibetan Herbin G. A. and Robins P. A. (1969) Patterns of variation and Plateau) and Holzmaar (Germany). Org. Geochem. 39, development in leaf wax alkanes. Phytochemistry 8, 1985–1998. 711–729. Huang Y., Lockheart M. J., Collister J. W. and Eglinton G. (1995) Nichols J. E., Booth R. K., Jackson S. T., Pendall E. G. and Huang Molecular and isotopic biogeochemistry of the Miocene Clarkia Y. (2006) Paleohydrologic reconstruction based on n-alkane Formation: hydrocarbons and alcohols. Org. Geochem. 23, distributions in ombrotrophic peat. Org. Geochem. 37, 1505– 785–801. 1513. Jansen B., Nierop K. G. J., Hageman J. A., Cleef A. M. and Nichols J., Booth R. K., Jackson S. T., Pendall E. G. and Huang Verstraten J. M. (2006) The straight-chain lipid biomarker Y. (2010) Differential hydrogen isotopic ratios of Sphagnum composition of plant species responsible for the dominant and vascular plant biomarkers in ombrotrophic peatlands as a biomass production along two altitudinal transects in the quantitative proxy for precipitation – evaporation balance. Ecuadorian Andes. Org. Geochem. 37, 1514–1536. Geochim. Cosmochim. Acta 74, 1407–1416. Jansen B., van Loon E. E., Hooghiemstra H. and Verstraten J. M. Nott C. J., Xie S., Avsejs L. A., Maddy D., Chambers F. M. and (2010) Improved reconstruction of palaeo-environments Evershed R. P. (2000) n-Alkane distributions in ombrotrophic through unravelling of preserved vegetation biomarker pat- mires as indicators of vegetation change related to climatic terns. Palaeogeogr. Palaeoclimatol. Palaeoecol. 285, 119–130. variation. Org. Geochem. 31, 231–235. Jetter R., Kunst L. and Samuels A. L. (2006) Composition of plant Osborne R., Salatino M. L. F. and Salatino A. (1989) Alkanes of cuticular waxes. In Biology of the Plant Cuticle (eds. M. foliar epicuticular waxes of the genus Encephalartos. Phyto- Riederer and C. Mu¨ller). Blackwell Publishing, Oxford. chemistry 28, 3027–3030. Kawamura K. and Ishiwatari R. (1984) Fatty acid geochemistry of Osborne R., Salatino A., Salatino M. L. F., Sekiya C. M. and a 200 m sediment core from Lake Biwa, Japan. Early diagenesis Torres M. V. (1993) Alkanes of foliar epicuticular waxes from and paleoenvironmental information. Geochim. Cosmochim. five cycad genera in the Zamiaceae. Phytochemistry 33, 607– Acta 48, 251–266. 609. Kawamura K., Ishimura Y. and Yamazaki K. (2003) Four years’ Otto A., Simoneit B. R. T. and Rember W. C. (2005) Conifer and observations of terrestrial lipid class compounds in marine angiosperm biomarkers in clay sediments and fossil plants from aerosols from the western North Pacific. Global Biogeochem. the Miocene Clarkia Formation, Idaho, USA. Org. Geochem. Cycles 17, 1–19. 36, 907–922. R.T. Bush, F.A. McInerney / Geochimica et Cosmochimica Acta 117 (2013) 161–179 179

Pancost R. D., Baas M., van Geel B. and Damste´ J. S. S. (2002) Salasoo I. (1987c) Epicuticular wax hydrocarbons of Ericaceae in Biomarkers as proxies for plant inputs to peat: an example from Germany. Z. Naturforsch. 47c, 499–501. sub-boreal ombrotrophic bog. Org. Geochem. 33, 675–690. Schefuß E., Ratmeyer V., Stuut J.-B. W., Jansen J. H. F. and Peters K. E., Walters C. C. and Moldowan J. M. (2005) The Damste´ J. S. S. (2003) Carbon isotope analyses of n-alkanes in Biomarker Guide. Cambridge University Press, Cambridge, dust from the lower atmosphere over the central eastern UK. Atlantic. Geochim. Cosmochim. Acta 67, 1757–1767. Piasentier E., Bovolenta S. and Malossini F. (2000) The n-alkane Schwark L., Zink K. and Lechterbeck J. (2002) Reconstruction of concentrations in buds and leaves of browsed broadleaf trees. J. postglacial to early Holocene vegetation history in terres- Agric. Sci. 135, 311–320. trial.Central Europe via cuticular lipid biomarkers and pollen Post-Beittenmiller D. (1996) Biochemistry and molecular biology records from lake sediments. Geology 30, 463–466. of wax production in plants. Ann. Rev. Plant Physiol. Plant Shepherd T. and Griffiths D. W. (2006) The effects of stress on Mol. Biol. 47, 405–430. plant cuticular waxes. New Phytol. 171, 469–499. Quenea K., Derenne S., Largeau C., Rumpel C. and Mariotti A. Smith F. A., Wing S. L. and Freeman K. H. (2007) Magnitude of (2004) Variation in lipid relative abundance and composition the carbon isotope excursion at the Paleocene–Eocene thermal among different particle size fractions of a forest soil. Org. maximum: the role of plant community change. Earth Planet. Geochem. 35, 1355–1370. Sci. Lett. 262, 50–65. Rinna J., Gu¨nter U., Hinrichs K.-U., Mangelsdorf K., van der Strańsky´ K., Streibl M. and Herout V. (1967) On natural waxes. Smissen J. H. and Rullko¨tter J. (1999) Temperature-related VI. Distribution of wax hydrocarbons in plants at different molecular proxies: degree of alkenone unsaturation and average evolutionary levels. Coll. Czechoslovak Chem. Commun. 32, chain length of n-alkanes. In Proceedings of the Sixteenth 3213–3220. Annual Pacific Climate Workshop, May 24–27, 1999 (eds. J. G. Tipple B. J., Berke M. A., Doman C. E., Khachaturyan S. and West and L. Buffaloe). The Wrigley Institute for Environmental Ehleringer J. R. (2013) Leaf-wax n-alkanes record the plant- Studies, Two Harbors, Santa Catalina Island, California. `ıwater environment at leaf flush. Proc. Natl. Acad. Sci. 110, Rommerskirchen F., Eglinton G., Dupont L., Gu¨nter U., Wenzel 2659–2664. C. and Rullko¨tter J. (2003) A north to south transect of Tu T. T. N., Egasse C., Zeller B. and Derenne S. (2007) Holocene southeast Atlantic continental margin sediments: Chemotaxonomical investigations of fossil and extant beeches. relationship between aerosol transport and compound-specific I. Leaf lipids from the extant Fagus sylvatica L. C. R. Palevol 6, d13C land plant biomarker and pollen records. Geochem. 451-461. Geophys. Geosyst. 4, 1101–1128. Vioque J., Pastor J. and Vioque E. (1994) Leaf wax alkanes in the Rommerskirchen F., Plader A., Eglinton G., Chikaraishi Y. and genus Coincya. Phytochem. 36, 349–352. Rullko¨tter J. (2006) Chemotaxonomic significance of distribu- Vogts A., Moossen H., Rommerskirchen F. and Rullko¨tter J. tion and stable carbon isotopic composition of long-chain (2009) Distribution patterns and stable carbon isotopic com-

alkanes and alkan-1-ols in C4 grass waxes. Org. Geochem. 37, position of alkanes and alkan-1-ols from plant waxes of African 1303–1332. rain forest and savanna C3 species. Org. Geochem. 40, 1037– Sachse D., Radke J. and Gleixner G. (2004) Hydrogen isotope 1054. ratios of recent lacustrine sedimentary n-alkanes record modern Vogts A., Schefuß E., Badewein T. and Rullko¨tter J. (2012) n- climate variability. Geochim. Cosmochim. Acta 68, 4877–4889. Alkane parameters from a deep sea sediment transect off Sachse D., Radke J. and Gleixner G. (2006) dD values of individual southwest Africa reflect continental vegetation and climate n-alkanes from terrestrial plants along a climatic gradient– conditions. Org. Geochem. 47, 109–119. Implications for the sedimentary biomarker record. Org. Vonk J. E. and Gustafsson O¨ . (2009) Calibrating n-alkane Geochem. 37, 469–483. Sphagnum proxies in sub-Arctic Scandinavia. Org. Geochem. Sachse D., Kahmen A. and Gleixner G. (2009) Significant seasonal 40, 1085–1090. variation in the hydrogen isotopic composition of leaf-wax Vonk J. E., van Dongen B. E. and Gustafsson O¨ . (2008) Lipid lipids for two deciduous tree ecosystems (Fagus sylvatica and biomarker investigation of the origin and diagenetic state of Acer pseudoplatanus). Org. Geochem. 40, 732–742. sub-arctic terrestrial organic matter presently exported into the Sachse D., Gleixner G., Wilkes H. and Kahmen A. (2010) Leaf wax northtern Bothnian Bay. Mar. Chem. 112, 1–10. n-alkane dD values of field-grown barley reflect leaf water dD Wakeham S. G. (1976) A comparative survey of petroleum values at the time of leaf formation. Geochim. Cosmochim. Acta hydrocarbons in lake sediments. Mar. Pollut. Bull. 7, 206–211. 74, 6741–6750. Zech M., Buggle B., Leiber K., Markovic´ S., Glaser B., Hambach Sachse D., Billault I., Bowen G. J., Chikaraishi Y., Dawson T. E., U., Huwe B., Stevens T., Su¨megi P., Wiesenberg G. and Zo¨ller Feakins S. J., Freeman K. H., Magill C. R., McInerney F. A., L. (2009) Reconstructing Quaternary vegetation history in the van der Meer M. T. J., Polissar P., Robins R. J., Sachs J. P., Carpathian Basin, SE Europe, using n-alkane biomarkers as Schmidt H.-L., Sessions A. L., White J. W. C., West J. B. and molecular fossils: projects and possible solutions, potential and Kahmen A. (2012) Molecular paleohydrology: interpreting the limitations. Quatern. Sci. J. 58, 148–155. hydrogen-isotopic composition of lipid biomarkers from pho- Zech M., Krause T., Meszner S., and Faust D. (2013) Incorrect tosynthesizing organisms. Annu. Rev. Earth Planet. Sci. 40, when uncorrected: Reconstructing vegetation history using n- 221–249. alkane biomarkers in loess-paleosol sequences–A case study Salasoo I. (1981) Alkane distribution in epicuticular wax of some from the Saxonian loess region, Germany. Quaternary Int. 296, evergreen Ericaceae. Can. J. Bot. 59, 1189–1191. 108-116. Salasoo I. (1983) Alkane distribution in epicuticular wax of Zhang Z., Zhao M., Eglinton G., Lu H. and Huang C.-Y. (2006) Epacridaceae. Phytochem. 22, 937–942. Leaf wax lipids as paleovegetational and paleoenvironmental Salasoo I. (1987a) Alkane distribution in epicuticular wax of some proxies for the Chinese Loess Plateau over the last 170 kyr. heath plants in Norway. Biochem. Syst. Ecol. 15, 663–665. Quatern. Sci. Rev. 25, 575–594. Salasoo I. (1987b) Epicuticular wax alkanes of some heath plants in central Alaska. Biochem. Syst. Ecol. 15, 105–107. Associate editor: Peter Hernes