Biomarker proxies for reconstructing Quaternary climate and environmental change

To reconstruct past environmental changes, a range of indirect or proxy approaches can be applied to Quaternary archives. Here, we review the complementary and novel insights that have been provided by the analysis of chemical fossils (biomarkers). Biomarkers have a biological source that can be highly specific (e.g. produced by a small group of organisms) or more general. We show that biomarkers are able to quantify key climate variables (particularly water and air temperature) and can provide qualitative evidence for changes in hydrology, vegetation, human–environment interactions and biogeochemical cycling. In many settings, biomarker proxies provide the opportunity to simultaneously reconstruct multiple climate or environmental variables, alongside complementary and long‐established approaches to palaeoenvironmental reconstruction. Multi‐proxy studies have provided rich sets of data to explore both the drivers and impacts of palaeoenvironmental change. As new biomarker proxies continue to be developed and refined, there is further potential to answer emerging questions for Quaternary science and environmental change.


Introduction
To reconstruct past environmental and climate changes, indirect physical, chemical or biological signals of environmental variables ('proxies') are recovered from a range of archives (e.g.marine and lake sediments, ice cores, speleothems, peatlands).Biomarker proxies are molecular or chemical fossils with a biological origin (Eglinton and Calvin, 1967), which can be recovered, analysed and identified from palaeoenvironmental archives (Peters et al., 2005).Biomarkers have emerged as valuable parts of the Quaternary science toolkit, due to both quantitative and qualitative insights into past environmental changes and because multiple biomarkers (and thus multiple environmental signals) are simultaneously recovered from single samples.
Biomarkers can either be very specific in terms of their environmental signal or biological source (e.g.individual highly branched isoprenoids indicative of sea-ice diatoms), or be more general indicators (e.g.mixtures of n-alkanes derived from higher plants) (Figure 1).A key strength of biomarker analysis is that biomarkers from multiple settings can be found in a single sediment sequence, since terrestrial biomarkers (from bedrock, soils or plants) may be transported by wind, rivers or ice into lakes, wetlands, caves or marine environments, allowing both the transport process and changes in different environments to be explored (e.g.Jaffé et al., 2001;Ngugi et al., 2017;Müller et al., 2018).Biomarker transport can also be a disadvantage: advection or bioturbation may influence how biomarkers are incorpo-rated into the sediments and can even lead to age-offsets between different proxies (e.g.Ohkouchi et al., 2002).As organic molecules, biomarkers are subject to degradation processes during transport and deposition (e.g.Madureira et al., 1997;Wakeham et al., 1997;Thomas et al., 2001).However, different classes of organic compounds have varying rates of degradation (Arndt et al., 2013).Some of the most widely applied biomarkers are those which are relatively resistant to alteration (e.g.plant waxes), or where (rapid) transformation of lipids or pigments found in living biomass leaves behind a recognizable chemical signal so that the source organisms or formation processes can be determined (e.g.Harris et al., 1996;Pitcher et al., 2009).Biomarkers may be particularly useful in environments where other proxies (e.g.plant macrofossils) are degraded but their chemical remains can be found (e.g.Ronkainen et al., 2015).
A valuable property of biomarkers is that they can be isolated from the original archive so that isotope analysis can be undertaken on individual components of organic matter with a known origin.This 'compound-specific isotope analysis' (CSIA) contrasts with the analysis of bulk samples, where changing isotope ratios could reflect varying contributions of different organic sources through time or space, as well as environmental controls over the contributing isotopic signals (e.g.Holtvoeth et al., 2019;McClymont et al., 2022).By knowing the origin of the biomarker, the relative impact of biological and environmental controls on stable isotope ratios can be determined (Sachse et al., 2012;Holtvoeth et al., 2019).CSIA has enabled, for example, separation of the contributions of C 3 and C 4 plants and isolation of hydrological controls over plant wax deuterium/ hydrogen isotopes (see 'Reconstructing vegetation and hydrological change using compound-specific stable isotope analysis').
Our aim in this review is to provide an accessible introduction to the wide range of biomarker applications in Quaternary science.Detailed reviews are also available on both biomarker synthesis and proxy development in marine sediments (Rosell-Melé, McClymont, 2007), lake sediments (Castañeda and Schouten, 2011), peatlands (Naafs et al., 2019), speleothems (Blyth et al., 2016;Meckler et al., 2021) and geoarchaeology (Dubois and Jacob, 2016).Here, we explore a range of studies that have applied biomarker proxies and outline the novel and complementary contributions biomarkers have made to palaeoenvironmental reconstructions across a wide range of geographical regions, timescales and environments.The review was conducted using methodical keyword literature searches of the Web of Science and Google Scholar databases.The searches returned thousands of results; therefore, the scope of this review precludes citations of all relevant studies.To address our aim of providing an accessible overview of biomarkers and their applications for all Quaternary scientists, we have prioritized the inclusion of initial foundation studies alongside a diversity of examples that span different timescales, sedimentary archives, geographical locations and topics of Quaternary science.Since some biomarker proxies have been applied to multiple archives but reconstruct similar environmental variables (e.g.temperature, salinity), the review is structured according to those variables or research questions, and archive-or proxy-specific considerations are provided.Finally, we reflect on recent developments in biomarker research and consider their future potential in Quaternary science.

An overview of biomarker laboratory methods
Biomarkers used in Quaternary studies include waterinsoluble lipids, photosynthetic pigments and macromolecules including lignin.Biomarkers are often present in very low (trace) concentrations in environmental samples (milligrams or nanograms per gram of material) and may be components of a complex matrix of organic and minerogenic materials.Isolating the biomarkers of interest requires methods that maximize recovery and minimize contamination.As multiple biomarkers are recovered simultaneously, a diverse range of environmental signals can be attained from a single sample.
Lipids and pigments are extracted from environmental or archaeological samples by using a range of organic solvents and approaches, tailored to the chemical properties of the compound(s) of interest.Ultra-sonication, microwave or accelerated solvent extraction methods are most commonly used but may have different efficiencies depending on sample size and composition (e.g.Kornilova and Rosell-Melé, 2003;Nichols, 2010;Kehelpannala et al., 2020;Manley et al., 2020).Lipid biomarkers are commonly extracted with dichloromethane and methanol in a ratio aligned with the expected polarity of the target marker, whereas pigments are typically extracted using acetone (e.g.Chen et al., 2001) or a mixture of acetone, methanol and water (Leavitt and Hodgson, 2001).Pigment extractions can include soaking overnight at low temperatures (e.g.20 °C) to minimize degradation (Jeffrey et al., 1997).Care is needed, because organic solvents will also extract unwanted compounds and add them to the extract, particularly plasticizers but also oils from the skin/hair of researchers handling the materials (e.g.Blyth et al., 2006).As a result, subsampling cores or materials using metal spatulas, storing samples and extracts in glass jars or high-quality (low contaminant) bags, and using foil to separate samples from plastic bags or lids are effective strategies for minimizing contamination, alongside using laboratory personal protective equipment (e.g.Nichols, 2010).Inclusion of blanks during sample processing allows for contamination to be detected, monitored and isolated (Blyth et al., 2016).Water can also interfere with lipid extraction efficiency and subsequent cleanup steps, and encourages oxidative degradation; the best approach is to freeze-dry samples (McClymont et al., 2007;Nichols, 2010).
A common approach in palaeoenvironmental research is to recover multiple lipid biomarkers in a single extraction procedure to generate an 'extract' (Kornilova and Rosell-Melé, 2003;Nichols, 2010).The extract may then be separated into classes of compounds according to their chemistry (e.g.polarity, pH) to isolate the target biomarkers or to remove interfering compounds (Nichols, 2010).Biomarkers are then analysed using liquid (LC) or gas chromatography (GC), whereby a prepared sample is introduced to a capillary column and transferred to a detector by a flow of liquid or gas (Peters et al. 2005).Non-extractable material (e.g.lignins) can be introduced by pyrolysis, whereby high temperatures are used to split the large, refractory, molecules into diagnostic fragments (White et al., 2004).The capillary column (usually 0.20-0.25 mm internal diameter) is coated with an internal film called the stationary phase, the chemistry of which determines how compounds are retained and released according to their chemical properties as they travel through the column.The result is a chromatogram of individual compounds separated by their chemical interaction with the column (Figure 2).
Biomarker identification usually involves the separated individual compounds being transferred directly to a mass spectrometer (LC-MS, GC-MS), which ionizes and fragments them into characteristic patterns (Peters et al., 2005).Semiquantitative analysis can be achieved by adding internal standards of known mass during the extraction steps, or a calibration curve will be derived using external standards of varying concentrations to enable absolute quantification (e.g.McGowan, 2013).However, some analysis remains qualitative where internal standards are not feasible (e.g.McClymont et al., 2011).Ratios between different compounds may be more appropriate for characterizing changing biomarker distributions; for several biomarkers these ratios are defined as indices which are specifically linked to, or calibrated against, environmental variables (Tables 1 and 2).
Finally, the separation of organic matter also allows for CSIA.Not all samples or compounds are suitable: individual biomarkers need to meet higher detection limits than for GC or LC, and there needs to be excellent baseline separation between peaks.For compound-specific 14 C analysis, GC or LC techniques can be used to separate and then collect individual compounds or classes of compounds for subsequent analysis (Eglinton et al., 1996;Yamane et al., 2014;Sun et al., 2020).

Biological functions of biomarkers
In this section we have selected examples to introduce the biological function of biomarkers and the mechanistic principles behind their palaeoenvironmental proxy applications.The biological function of biomarkers varies between different classes of compounds (Peters et al., 2005;Bianchi and Canuel, 2011;Killops and Killops, 2013).Most lipid biomarkers used within Quaternary research can be classified as leaf wax or cell membrane lipids.Leaf wax lipids, such as n-alkanoic acids and n-alkanes, are synthesized by vegetation to act as waterproof protective barriers against the external environment and to control evaporative water loss and gas exchange (Eglinton and Hamilton, 1967;Post-Beittenmiller, 1996;Jetter et al., 2006).The chain length of BIOMARKER PROXY REVIEW leaf wax molecules varies between different plant species and hydrological conditions: aquatic (terrestrial) species are characterized by shorter (longer) chain lengths since they are adapted to wetter (drier) conditions (Cranwell et al., 1987;Ficken et al., 2000;Schefuß et al., 2003;Table 2).Biochemical responses to environmental conditions can occur at fine scales, which should be considered during interpretation of the sedimentary record.For example, n-alkane chain lengths (Ronkainen et al., 2013) or concentrations (Huang et al., 2011) have been shown to differ between the leaves and roots of wetland species (Ronkainen et al., 2013;Andersson et al., 2011), and both humidity and the timing of leaf growth can impact n-alkane distributions even within single plants (e.g.Sachse et al., 2010;Eley and Hren, 2018).There is also evidence for loss and transformation of some n-alkyl components within soils, although the dominant chain lengths tend to be maintained with depth (Thomas et al., 2001).
Cell membrane lipids are synthesized by a range of organisms including fungi, algae, plants and animals (such as sterols, e.g.Volkman, 1986), archaea (including isoprenoidal glycerol dialkyl glycerol tetraethers [isoGDGTs], e.g.Nishihara andKoga, 1987, Sinninghe Damsté et al., 2000) Table 2. Examples of lipid ratios and compounds used to identify differences in vegetation source and environmental conditions.

Incomplete combustion of organic matter
Proxy for vegetation burning.Some alkylated PAHs are also formed during thermal maturation and petrogenic processes, but ratios have been applied to distinguish between (non)pyrogenic sources and identify vegetation type Ramdahl (1983) and eubacteria (such as hopanoids, e.g.Innes et al. 1997;Ourisson et al. 1979).Membrane lipids are structural components of cells that provide a stable controlled environment for biogeochemical reactions.Cell membrane lipids regulate the fluidity (or permeability) of the cell membrane by altering structural features such as chain lengths, the placement of unsaturated (double) bonds and cyclic rings (Peters et al., 2005;Bianchi and Canuel;2011;Killops and Killops, 2013; Figure 3).For example, temperature changes are expressed by the number and position of methyl groups of branched GDGTs (brGDGTs; Weijers et al., 2007) and the number of cyclopentane moieties of isoGDGTs (De Rosa et al., 1980) (Figure 3; see next section).The primary functions of some lipids remain unknown or poorly understood.For example, alkenones, synthesized by phytoplankton (Theroux et al., 2010), were originally considered to be fluidity-influencing membrane lipids (e.g.Brassell et al., 1986); however, more recent studies have demonstrated that they more probably contribute to energy storage and regulate properties such as melting point and therefore ease of lipid catabolism (e.g.Epstein et al., 2001;Bakku et al., 2018).Regardless of their specific function, differences in alkenone chain lengths and the degree of unsaturation (number of double bonds) can be used to reconstruct palaeotemperature (e.g.Brassell et al., 1986;Figure 3; see 'Ocean and lake temperature reconstructions').Some other types of biomarkers of interest to Quaternary scientists are transformation products that reflect environmental processes.For example, some polyaromatic hydrocarbons (PAHs) and monosaccharide anhydrides (MAs) are produced during combustion of organic matter and can therefore be used to reconstruct fire histories (see 'Biomarkers of burning and agricultural activity as indirect indicators of human activity').
Pigments can be relatively general biomarkers of photosynthetic processes (e.g.chlorophyll a/b/c and ββ-carotene are general productivity markers) or highly specific (e.g.alloxanthin is only found in cryptophytes; reviewed by McGowan, 2007).Pigment functions also vary: chlorophylls are active sites of photosynthesis, providing energy for the cell, whereas carotenoids can also help absorb light for photosynthesis (Jeffrey et al., 1997) or help protect cells from ultraviolet (UV) exposure (e.g.scytonemin; McGowan, 2007).The stability of pigments is dependent on specific chemistry, the environment and presence of photoprotection (Leavitt, 1993;Cuddington and Leavitt, 1999).Some pigments are susceptible to oxidative or UV degradation, and even in environments with good preservation there can be as much as 95% degradation in the water column before sedimentation (McGowan, 2007).Pigment analysis is thus often most effective in environments where preservation is facilitated by, for example, anoxic or low light conditions (e.g.Hodgson et al., 2005).Where degradation allows characteristic fragments of the original pigment to be identified, valuable information can be recovered.For example, chlorins represent the preserved central ring structure of the original chlorophyll and are frequently selected as marine productivity biomarkers over glacial-interglacial timescales (Harris et al., 1996) (see 'Reconstructing biological productivity in lakes and the oceans').

Quantifying amplitudes and rates of past temperature change
Air, water and soil temperatures are important for detailing climate system response to radiative forcing, including global climate sensitivity (Masson-Delmotte et al., 2021).Temperatures trace heat transfers through ocean/atmosphere circulation systems and can be informative of local conditions which may influence ecosystems.Quantification of past temperature change has been a key achievement for biomarker proxies and continues to be a frontier of biomarker proxy development.Here, we first outline insights gained from marine and lacustrine settings, before discussing emerging terrestrial records from soils, peats and speleothems.

Ocean and lake temperature reconstructions
An early biomarker proxy success was the recognition that some aquatic organisms change their cell membrane chemistry in response to water temperature, and that these signals were detectable in sediments (Brassell et al., 1986; Figure 3).Multiple biomarker temperature proxies have subsequently been developed (Table 1).Biomarker temperature indices describe distributions of lipids produced by selected photosynthesizing haptophyte algae (alkenone-derived U K 37 ′ and U K 38Me ′ indices; Prahl and Wakeham, 1987;Novak et al., 2022), ammonia-oxidizing Thaumarchaeota (isoGDGTderived TEX 86 index; Schouten et al., 2022), eustigmatophyte algae (long-chain alkyl diol-derived LDI; Rampen et al., 2012) and bacteria (hydroxy fatty acid-derived RAN 13 index and brGDGT-derived MBT′ 5Me index; De Jonge et al., 2014;Yang et al., 2020).As each proxy has different source organisms and controls (Table 1), there is potential to generate detailed water temperature reconstructions which might include seasonality or temperature profiles with water depth.Both the U K 37 ′ and TEX 86 proxies have reconstructed temperatures through the Quaternary and beyond (e.g.Herbert et al., 2010); more recently developed proxies have tended to focus on the Holocene or the last glacial cycle (e.g.Powers et al., 2005;Warnock et al., 2018;Yang et al., 2020).
Biomarker water temperature proxies are calibrated using field sampling, laboratory culture experiments and sediment core-tops (Table 1).The accuracy and precision of the temperature proxies vary, especially at the upper and lower ends of the calibrations or close to detection limits, and not all proxies are found in all settings.Many of the proxies are calibrated to mean annual surface water temperature (Table 1), but if the producers have preferred seasons or water depths, a seasonal or subsurface temperature signal may be reconstructed (D'Andrea et al., 2005(D'Andrea et al., , 2011;;Jaeschke et al., 2017;Tierney and Tingley, 2018;Inglis and Tierney, 2020;Theroux et al., 2020;Spencer-Jones et al., 2021).Although marine biomarkers have global calibrations (Table 1), there can also be local controls over the biomarker-temperature relationship in all aquatic settings (e.g.salinity, sea/lake ice cover, lake size).In some settings a regional temperature calibration may be more appropriate (Table 1) (e.g.Bendle et al., 2005;De Jonge et al., 2014;D'Andrea et al., 2016;Loomis et al., 2014;Longo et al., 2016;De Bar et al., 2020;Sinninghe Damsté et al., 2022;Yao et al., 2022).
A key impact of marine sea surface temperature (SST) biomarker proxies has been the generation of quantitative data to calculate amplitudes and rates of change, climate response to changing CO 2 , and to facilitate data-model comparisons (e.g.Brassell et al., 1986;MARGO Project Members, 2005;Martrat et al., 2007;Schmittner et al., 2011;Capron et al., 2017;Tierney et al., 2020).Relatively strong mid-and high-latitude SST responses to glacial-interglacial cycles have been demonstrated (Martrat et al., 2007;Naafs et al., 2013), but tropical cooling has also been reconstructed during glacials (MARGO Project Members, 2005;Herbert et al., 2010;McClymont et al., 2013).U K 37 ′ records have shown that there are regional and temporal differences in the amplitudes of interglacial warming (MARGO Project Members, 2005; Past Interglacials Working Group of PAGES, 2016) and that early ocean cooling preceded the evolution of 100-ka glacial-interglacial cycles during the mid-Pleistocene transition (McClymont et al., 2013).Recent calibration of the U K 38Me ′ index shows potential to extend the upper linear calibration limit of the U K 37 ′ proxy to ~30 °C (Novak et al., 2022), reducing the reconstructed uncertainties at high SSTs (Table 1) and enabling improved reconstructions of inter-glacial warmth and glacial-interglacial variability in the low latitudes.
Differences in absolute SSTs from U K 37 ′ and TEX 86 or LDI reconstructions from the same sediment sequences have revealed circulation changes on a range of timescales (Figure 4).In the first TEX 86 reconstruction spanning the last deglaciation from the South China Sea, SSTs aligned well with millennial-scale variability in Hulu Cave stalagmite δ 18 O, but exceeded and had a different trend to the U K 37 ′ SSTs, which may in part be explained by different seasons of production (Shintani et al., 2011).In low-latitude upwelling systems, higher U K 37 ′ (surface) and lower TEX 86 (subsurface) temperatures have enabled reconstructions of varying upwelling intensity spanning millennial to million-year timescales (e.g.McClymont et al., 2012;De Bar et al., 2018;Petrick et al., 2018;Erdem et al., 2021).Glacial-interglacial migrations in the latitude of the Subtropical Front in the southern hemisphere have been determined by combining U K 37 ′ and TEX 86 data (Cartagena-Sierra et al., 2021), and seasonally driven offsets between U K 37 ′, TEX 86 and LDI temperatures identified variable Leeuwin Current strength offshore southeast Australia over the last ~135 ka (Lopes dos Santos et al., 2013a).Although less widely applied, the LDI has isolated Baltic Sea cooling related to the 8.2-ka event, followed by a Holocene Thermal Maximum, and late Holocene cooling with sea-ice expansion (Warnock et al., 2018).On much shorter timescales, an 'Atlantification' of waters in the Fram Strait through the 20th century was detected using U K 37 ′ and TEX 86 (Tesi et al., 2021;Figure 4).Here, a multi-biomarker approach, with 5-10-year resolution, enabled interactions between sea ice, ocean mixing and heat transfer to be better determined than by using the short instrumental record alone.
Lake temperature reconstructions provide valuable climate indicators for continental climate change.Early TEX 86 records generated new constraints on temperature change in Africa: an ~2 °C increase in Lake Malawi surface water temperature occurred during the last ~100 years which exceeded variability in the preceding ~600 years (Powers et al., 2005); coherence between Lake Victoria warming/cooling and rainfall occurred over the last ~14 000 years (Berke et al., 2012a); and both longterm and abrupt temperature changes in Lake Tanganyika were linked to Indian Ocean SSTs across the last deglaciation (Tierney et al., 2008).However, local or regional influences over the biomarker-temperature relationships include lake size and depth (for TEX 86 ; Sinninghe Damsté et al., 2022), salinity or alkalinity (for MBT′ 5Me and alkenones; Pearson et al., 2008;De Jonge et al., 2014;Song et al., 2016;Plancq et al., 2018), nutrient availability (Toney et al., 2010), and inputs of soils containing the same compounds (e.g.Loomis et al., 2012;De Jonge et al., 2015;Russell et al., 2018).GDGT inputs from methanogens and other archaea can also complicate TEX 86 reconstructions: at Lake Challa (Africa) reliable temperature reconstructions using lacustrine GDGTs were only possible between 25 and 13 ka, but not in the Holocene section (Sinninghe Damsté et al., 2012).
The brGDGT proxy MBT′ 5Me (de Jonge et al., 2014) has been used to reconstruct millennial-and centennial-scale variations in lake temperature, which align with stadial and interstadial events in the Iberian Peninsula (Rodrigo-Gámiz et al., 2022).Although local conditions prevented application of the MBT′ 5Me index to an Icelandic lake, the combined analysis of brGDGT distributions and U K 37 ′ data enabled quantification of temperature change through the Holocene which could be directly compared to reconstructed and modelled ice cap change (Harning et al., 2020).Having quantified early Holocene warmth, the loss of the local ice cap by ~2050 CE was predicted (Harning et al., 2020).A challenge for brGDGT reconstructions is that the calibration uncertainties (up to ~5 °C; Table 1) are of similar magnitude to some reconstructed Quaternary temperature changes.The application of MBT′ 5Me can be Figure 3.A selection of palaeotemperature biomarkers, detailing the different chemical properties that can be used to identify specific markers and their relationships to biological and environmental variables.(a) TEX 86 (TetraEther indeX of tetraethers consisting of 86 carbon atoms) temperature proxy is calculated using the relative distributions of isoGDGTs (iso-GDGT-1, iso-GDGT-2 and iso-GDGT-3) and the crenarchaeol regioisomer (Schouten et al., 2002).Blue circles highlight the number of cyclopentane moieties, and the yellow circle highlights the presence of a cyclohexane ring.(b) MBT′ 5Me (Methylation of Branched Tetraethers using the 5-methyl isomers) temperature proxy in soils is calculated using the relative distributions of 5-methyl brGDGT (de Jonge et al., 2014).Blue circles highlight the presence and number of cyclopentane moieties and green circles highlight the presence and number of methyl groups in the α and/or ω-5 position.(c) U K 37 ′ temperature proxy in freshwater and marine environments is calculated using the relative distributions of the di-and tri-unsaturated alkenone distributions (Prahl and Wakeham, 1987).The chain lengths of the two alkenones are the same (C 37 = 37 carbon atoms), but the number of double bonds increases from two to three (highlighted by red circles).(d) LDI (long chain diol index) temperature proxy in freshwater and marine environments is calculated using the relative distributions of C 28 and C 30 1,13-and C 30 1,15-alykl diol distributions (Rampen et al., 2012;2014).Compounds vary in terms of chain lengths (C 28 = 28 carbons atoms and C 30 = 30 carbons atoms) and the location of the midchain alcohol group (C 13 or C 15 ; highlighted by the orange circles).(e) RAN 15 temperature proxy in soils is calculated using the ratio of anteiso to normal 3-hydroxy C 15 fatty acid (Wang et al., 2021a).Green circle highlights the methyl-substituent located on the antepenultimate carbon atom.(Lamb, 1977).[Color figure can be viewed at wileyonlinelibrary.com] complex since the full range of specific bacterial sources of brGDGTs is unknown: community sequencing of laboratory cultures, environmental samples, and micro-and mesocosm studies have identified Acidobacteria as brGDGT producers; however, they currently do not account for the full distributions of brGDGTs found in sedimentary samples (e.g.Weijers et al., 2010;Sinninghe Damsté et al., 2011, 2018;Martinez-Sosa and Tierney, 2019;De Jonge et al., 2021;Halamka et al., 2023).
The uncertainty surrounding the producer organisms (and whether they have changed through time), as well as limited high-latitude samples in global calibrations (Blaga et al., 2010;De Jonge et al., 2014;Naafs et al., 2017), has complicated the interpretation of Greenland lake data which did not align with other biomarker or macrofossil proxies (Kusch et al., 2019).In the high latitudes of the southern hemisphere, accounting for distinct brGDGT distributions at low temperatures enabled the production of a regional brGDGT calibration with reduced uncertainties; in turn, millennial-scale temperature changes were identified in an Antarctic lake core spanning the last ~4000 years (Foster et al., 2016).In East Africa, a regional MBT ′ 5Me calibration also reduced temperature reconstruction errors to <2.5 °C (Russell et al., 2018).Regional calibrations may therefore need to be considered where strong environmental impacts on lipid synthesis could occur.
Identification of key alkenone producers in North American, Greenland and Alaska lakes, with a preferred spring signal (e.g.D'Andrea, Huang, 2005;Toney et al., 2010;Wang et al., 2021a), offers the potential to quantify seasonal lake temperature change in the northern high latitudes.Centennial-scale late Holocene winter-spring lake temperature changes have been quantified in Iceland, showing a strong influence from SSTs (Richter et al., 2021).Holocene lake temperature changes linked to ice shelf configuration were reconstructed in northeast Greenland (Smith et al., 2023).With the recent development and calibration of the 3-hydroxy-fatty acid ratios in lakes (Table 1; e.g.Wang et al., 2021a) there is also the potential for new bacteria-derived temperature proxies to be generated, but downcore applications are not yet available.
In aquatic settings where there are inputs of organic matter from the continents, and where the same biomarkers are found onshore, it is important to assess and correct (or remove) temperature data which may incorporate a mixture of both marine and terrestrial inputs, since the two environments have different biomarker-temperature calibrations (e.g.De Jonge et al., 2015;Russell et al., 2018;Martínez-Sosa et al., 2021).For example, samples with high inputs of terrestrial brGDGTs can be flagged and removed using the BIT index (branched and isoprenoid tetraether index; Table 3 and Hopmans et al., 2004), whereas two separate calibrations may be applied if there is sedimentological evidence for a switch from marine to lake environments (Smith et al., 2023).Where a separation between aquatic and terrestrial lipids can be achieved, it is possible to generate terrestrial temperature records using lake/ marine sediments (e.g.Blaga et al., 2010;Watson et al., 2018; see subsection below).
Finally, on Quaternary timescales, there is potential for evolution to alter the biomarker-temperature relationship.Although the marine U K 37 ′-SST relationship appears robust to evolutionary events in alkenone producers (McClymont et al., 2005), a long-term (million year) warming in TEX 86 at Lake El'gygytgyn in the Russian Arctic was influenced by archaeal community changes as landscape evolution influenced biogeochemical cycling (Daniels et al., 2021).On shorter timescales, alkenone temperature indices in saline lakes can be impacted by shifts between the dominant haptophytes (Yao et al., 2022).For example, salinity-driven changes in the haptophyte assemblage in Lake Van, Turkey, are suggested to have complicated the U K 37 ′-temperature reconstructions for the oldest part of the record (~100-270 ka) (Randlett et al., 2014).

Temperatures reconstructed from soils, peats and speleothems
The calibration of biomarker proxies for continental temperatures using soils, peats and speleothems has been more challenging than for aquatic settings and remains an active area of development (e.g.Weijers et al., 2007;Naafs et al., 2017;Meckler et al., 2021).Quantified temperature data can provide a valuable backdrop to understand the rich environmental information recovered from the same archives (e.g.vegetation and hydrological change, human activity; see 'Reconstructing vegetation and hydrological change' and 'Sedimentary records of humans and animals in Quaternary landscapes').
The (acido)bacteria-produced brGDGTs, found in soils, peats and speleothems, have been explored as temperature proxies given their promise in aquatic settings ('Ocean and lake temperature reconstructions' above).The uncertainties in the brGDGT temperature calibrations for peat (~4.7 °C, Naafs et al., 2017) Yamamoto et al., 2016) make it difficult to reconstruct lowamplitude and potentially brief Holocene temperature fluctuations.In the low latitudes, regional calibrations have been developed which have lower uncertainties (Pérez-Angel et al., 2020), and loess/palaeosol sequences have required careful interpretation given unusual brGDGT distributions in semi-arid settings (Yang et al., 2014).Conversion of soil or peat temperatures to overlying air temperatures has also been challenging where there are differences between the two (Dearing Crampton-Flood et al., 2020).Nevertheless, in the Great Lakes region (North America) brGDGT-inferred soil/air temperatures from a lake core aligned with pollen-based temperature reconstructions associated with the Bølling-Allerød (B-A) warming, Younger Dryas cooling and Holocene warming (Watson et al., 2018).Importantly, the brGDGT analysis was able to advance understanding beyond pollen-based interpretations by showing that the multicentennial lag in warming compared to Northern Hemisphere temperature syntheses was due to the effects of continentality and regional influences of ice-sheet extent rather than a delayed vegetation response (Watson et al., 2018).Where soil-derived biomarkers have been transported to different depositional settings, there can be complexity in the signature if the source regions have changed over time: shifting sediment provenance of brGDGT distributions recovered offshore of the Amazon basin over the last deglaciation impacted the reconstructed absolute air temperature time-series, due to the increasing influence of colder, higher-elevation inputs from the Andes into the Holocene (Bendle et al., 2010).
In Asia, both isoGDGTs (TEX 86 ) and brGDGTs have been used in peat, loess and speleothems to explore the drivers and impacts of shifts in the summer monsoon.In peats, the combination of proxies for temperature and hydrology can be effective in considering their different drivers and the potential for (a)synchrony (e.g.Peterse et al., 2014;Wang et al., 2017).A 130 000-year loess-palaeosol sequence yielded high-resolution brGDGT temperature reconstructions: local insolation was the main driver of temperature change, but temperatures led brGDGT-inferred precipitation changes with a lag length which was linked to the intensity of northern hemisphere glaciation (Peterse et al., 2014).Rapid brGDGT temperature changes across the Younger Dryas and ~3.2 ka in Southeast China occurred synchronously with pollen assemblage changes over the last ~30 000 years in a peat sequence, and also showed asynchrony between temperature and precipitation proxies during the last deglaciation (Wang et al., 2017).A 4 °C increase in mean annual air temperature was recorded by speleothem-TEX 86 over the last deglaciation: the warming pre-dated Indian Summer Monsoon strengthening but was closely aligned with SST records (Huguet et al., 2018).A pattern of early Holocene warmth followed by cooling towards the present day has been recorded by brGDGTs in peats (NE China; Zheng et al., 2018) and using the more recently developed fatty acid RAN 15 index in a Chinese speleothem (Wang et al., 2018; Table 1).Given the challenges of recovering biomarkers from low organic carbon archives, and concerns about the relative influence of cave micro-environments on each record (Blyth et al., 2016;Baker et al., 2019), the recovery of both GDGTs and the C 15 and C 17 fatty acids from speleothems shows huge potential for generating new terrestrial records of cave or air temperature (e.g.Li et al., 2011;Blyth et al., 2016;Baker et al., 2019).As speleothems can also yield fatty acid, n-alkanol and n-alkan-2one distributions, interpreted to reflect changing soil microorganism responses to Holocene climate change (Xie et al., 2003;Kalpana et al., 2021), there is further potential to consider ecosystem response to temperature change (see also 'Reconstructing vegetation using biomarker distributions'), especially as analytical developments reduce sample sizes (e.g.Meckler et al., 2021).

Reconstructing vegetation and hydrological change
Palaeovegetation and palaeohydrology records provide insights into drivers of climate change that impact precipitation/ evaporation and terrestrial ecosystem response.Water availability is essential to the functioning of ecosystems and societies; therefore, long-term hydrological records also provide essential context for understanding changes in habitat and landcover, diets, agricultural practices, settlement dynamics and societal structures through the Quaternary.
Different vegetation types have characteristic biomarker distributions and stable isotope ratios reflecting their biosynthetic pathways and biological responses to environmental conditions (Table 2, 'Biological functions of biomarkers').
When the biological source of the biomarkers is well constrained, CSIA has enabled the varying biological and environmental influences over δ 13 C and δ 2 H to be disentangled.CSIA has thus emerged as a powerful tool for reconstructing both past vegetation change and palaeohydrology (Castañeda and Schouten, 2011;Diefendorf and Freimuth, 2017;Holvoeth et al., 2019;Inglis et al., 2022).

Reconstructing vegetation using biomarker distributions
Plant-derived lipids were among the first to be characterized (Eglinton and Hamilton, 1967), and remain among the most frequently applied biomarker tools owing to their prevalence in Quaternary sequences, their relative resilience to decay, ease of analysis, and the diversity of environmental information that they contain within their distributions and isotopic compositions.Lignin-derived compounds have also been targeted as relatively well-preserved plant remains (e.g.Castañeda et al., 2009b;reviewed in Jex et al., 2014).Biomarker vegetation reconstructions commonly use distributions of n-alkyl compounds such as n-alkanes, n-alkanols, n-alkanoic acids and wax esters, but may also draw upon sterols, phenols and more specific compounds (defined in Table 2).Biomarker vegetation reconstructions are usually made at the family rather than the species level, so the taxonomic detail is lower than with other vegetation proxies [pollen, plant macrofossils, and sedimentary ancient DNA (sedaDNA)].However, the relative resistance of n-alkyl compounds to decay has enabled vegetation reconstructions in samples with low levels of macro-and micro-fossil preservation, particularly in wetlands (e.g.McClymont et al., 2008a;Ronkainen et al., 2015).Biomarkers are also considered less susceptible to the long-range transport processes that can complicate pollen analyses due to the hydrodynamic properties of the leaves they are derived from (Schwark et al., 2002).
Complexity is introduced where some plants produce nalkane distributions that contain peaks in both longer and shorter chain lengths.For example, some Sphagnum species produce a dominant n-alkane chain length of C 23 , but also have elevated C 31 , which complicates the use of the C 23 /C 31 ratio as a Sphagnum indicator (e.g.Andersson et al., 2011;Bingham et al., 2010;Bush and McInerney, 2013; Table 2).However, the presence of the sphagnum acid product 4isopropenylphenol may offer a complementary assessment of the relative Sphagnum inputs to peat cores (e.g.Boon et al., 1986;McClymont et al., 2011).There may also be a bias caused by variable n-alkyl lipid production.For example, some conifer groups (e.g.Pinaceae) produce significantly less n-alkanes than broadleaf species, whereas others (e.g.Podocarpaceae) are similar (Diefendorf and Freimuth, 2017).As such, in catchments where pollen analyses indicate conifers as being the dominant vegetation type, biomarker interpretations should be part of a multi-proxy assessment: in northern Poland, this approach enabled subdecadal shifts in vegetation during the last deglaciation to be determined in detail (Aichner et al., 2018).
As different vegetation types have particular moisture preferences, plant biomarkers have been used to assess palaeohydrology by reconstructing the relative contributions of different vegetation types to sedimentary archives including lake sediments (e.g.Meyers, 2003;Castañeda et al., 2009b), marine sediments (Castañeda et al., 2009a), peats (e.g.Pancost et al., 2002;Ortiz et al., 2010;Zhou et al., 2010) and palaeosols (e.g.Zhang et al., 2006) (Table 2).Concurrent changes in the peatland C 23 /C 29 n-alkane ratio (Sphagnum/ vascular plants) and solar irradiance highlighted the sensitivity of north-east American hydroclimate to solar forcing, and its amplification by the Arctic/North Atlantic Oscillation since the mid-Holocene (Nichols and Huang, 2012).A key area of research has been the development of multiple records of vegetation change linked to changes in the Asian monsoon.Peatland aquatic/terrestrial vegetation reconstructions using nalkanes identified Holocene intensification of the Indian Summer Monsoon in the Garwhal Himalyas, and in turn, regional heterogeneity in mid-to late Holocene monsoonal conditions in the Indian sub-continent (Bhattacharya et al., 2021).Speleothem reconstructions of changing ecosystem dynamics have also been generated using a diverse suite of compounds, including n-alkanes (e.g.Xie et al., 2003;Blyth et al., 2007), sterols (e.g.Rousseau et al., 1995), fatty acids (e.g.Wang et al., 2019a) and lignin phenols (e.g.Blyth and Watson, 2009;Heidke et al., 2019).For example, in a Chinese speleothem, ratios of long-chain n-alkanes and n-alkan-2-ones (from terrestrial vegetation) to shorter chain compounds (from soil organisms) recorded vegetation changes during the Last Glacial Maximum which could be linked to fluctuations in North Atlantic SSTs during the last deglaciation (Xie et al., 2003).However, biomarker distributions (and other proxies) tend to be used as part of the evaluation of biological and/or environmental controls over compound-specific stable carbon and hydrogen isotope ratios, rather than in isolation (e.g.Castañeda et al., 2009aCastañeda et al., , 2009b)).

Reconstructing vegetation and hydrological change using compound-specific stable isotope analysis
For higher plant biomarkers, stable carbon isotope analysis (δ 13 C) of individual lipids provides a powerful tool to reconstruct past vegetation changes, because different photosynthetic pathways can be distinguished by their impact on plant tissue δ 13 C (Liu et al., 2022).Thus, n-alkane δ 13 C from trees and shrubs using the C 3 (Calvin-Benson) pathway is on average >10 ppm lower than in n-alkane δ 13 C from plants using the C 4 (Hatch-Slack) pathway, which are mainly tropical grasses (Castañeda et al., 2009a).A range of additional factors impact fractionation which may need to be considered in interpreting δ 13 C records, including moisture availability (for C 3 plants), ecological or physiological changes and past 13 CO 2 values (Diefendorf and Freimuth, 2016).A common nomenclature when presenting stable isotope ratios of individual lipids is δ 13 C lipid , where 'lipid' is the chain length or the name of the lipid which has been analysed.
The long-term reliability of the leaf wax δ 13 C vegetation proxy has been demonstrated through comparisons with pollen records from the late Pleistocene (e.g.Tierney et al., 2010;Huang et al., 2006).Mixing models have successfully used δ 13 C differences to reconstruct shifts in the relative abundance of C 3 and C 4 with the caveat that bias may also be introduced by variable n-alkyl lipid production ('Reconstructing vegetation using biomarker distributions'; Garcin et al., 2014).In tropical Africa, δ 13 C lipid records have reconstructed variable trees/shrubs (C 3 ) and grasses (C 4 ) extending back to the early Pleistocene from both lake and marine sediments (e.g.Castanẽda et al., 2007;Schefuß et al., 2003).In Lake Challa, Africa, δ 13 C analysis of the C 31 n-alkane (δ 13 C 31 ) reconstructed a vegetation transition from C 4 -dominated plants during the glacial period to a mix of C 3 / C 4 plants ~16.5 cal ka BP, which persisted during the Holocene and reflected the combined influences of increasing atmospheric CO 2 concentrations and increasing monsoon rainfall (Sinninghe Damsté et al., 2011).N-alkane, n-alkanol and δ 13 C 31 have recorded glacial-interglacial switches between steppe vegetation (C 3 ) and warm season grasses (C 4 ) at the Chinese loess plateau over the last 170 ka (Zhang et al., 2006).In Olduvai Gorge, orbitally paced δ 13 C 31 variations demonstrated rapid and large shifts between closed C 3 woodlands and more open C 4 grasslands ~1.9 Ma, challenging previous reconstructions of relatively stable ecosystems in the early Pleistocene (Magill et al., 2013).The ecosystem variations were probably linked to SST oscillations and monsoon strength, and provide a backdrop for the emergence and dispersal of Homo (Magill et al., 2013), as also suggested for more recent hominid migrations (e.g.Castañeda et al., 2009a).
By comparing n-alkane flux and δ 13 C signals across multiple glacial-interglacial timescales offshore of the Angola Basin, a decoupling between enhanced dust deposition ~900 ka and orbital variability in δ 13 C 31 revealed the different impacts of trade wind response to northern hemisphere ice-sheet growth (driving dust) and vegetation responses to regional SST changes (Schefuß et al., 2003).Lignin phenol and n-alkane distributions alongside n-alkane δ 13 C spanning the last 23 ka in Lake Malawi reconstructed millennial-scale variability in vegetation linked to wet conditions in south-east Africa, and a dominance of higher plant signals in bulk δ 13 C was confirmed (Castañeda et al., 2009b).However, caution is required where there may be mixed aquatic/terrestrial or local/regional inputs in the same archive: contributions of aquatic C 27 and C 29 nalkanes to a lake sediment resulted in different δ 13 C variations compared to the terrestrial leaf wax δ 13 C 31 in the same core (Liu et al., 2015); varying inputs of local and more widely sourced leaf waxes to an estuarine sequence were identified by different δ 13 C signals recorded depending upon the n-alkane chain length (Carr et al., 2015).
A powerful and direct proxy measurement of hydroclimate comes from δ 2 H signatures of lipids derived from plants and algae, which track the δ 2 H of their environmental water sources (reviewed by Sachse et al., 2012).D/H fractionation of meteoric water is influenced by temperature, precipitation source and amount, elevation and distance from the ocean, which results in a distinctive geographical pattern of lower δ 2 H precipitation at increasing latitude (e.g.Craig and Gordon, 1965;Bowen and Revenaugh, 2003).Several environmental and biological processes contribute to further D/H fractionation between the source water and the lipids and can complicate the interpretation of palaeohydrological δ 2 H lipid signatures (Sachse et al., 2012;Sessions, 2016;Huang and Meyers, 2018): higher plant δ 2 H lipid is influenced by factors such as humidity, evapotranspiration rates, light, vegetation assemblage and plant physiological differences (e.g.Smith and Freeman, 2006;Hou et al., 2008;Liu and Yang, 2008;Yang et al., 2009;Kahmen et al., 2013), whilst algal δ 2 H lipid is influenced by metabolic processes, growth rate and phase, nutrients and temperature (e.g.Schouten et al., 2006;Sachse and Sachs, 2008;Wolhow et al., 2009;Zhang et al., 2009).Salinity also influences source water δ 2 H and D/H fractionation of both plant and algal lipids, facilitating the application of δ 2 H lipid as a palaeosalinity proxy (discussed in the next section).
Care is needed to disentangle changes in n-alkane δ 2 H that are driven by biological fractionation or vegetation change rather than hydroclimate (e.g.Liu et al., 2006;Wang et al., 2013;Griepentrog, et al., 2019).This can be effectively achieved by reconstructing vegetation change using pollen, biomarker distributions, leaf wax δ 13 C or sedaDNA.At Meerfelder Maar, western Europe, the influences of vegetation change and hydroclimate were assessed using n-alkane distributions, pollen and n-alkane δ 2 H, demonstrating that cooler and wetter conditions were established ~2.8 ka BP (Rach et al., 2017).Contrasting late Holocene δ 2 H dinosterol hydroclimate reconstructions from paired lakes in the western tropical Pacific showcases the importance of multi-site and multi-proxy data to distinguish between climate and other limnological drivers of hydrological change (Maloney et al., 2022).By combining n-alkane and n-acid distributions with n-alkane δ 13 C and δ 2 H, both vegetation (δ 13 C 31 and δ 13 C 33 ) and precipitation (δ 2 H C29 ) were recorded and could be separated (Wang et al., 2013).Under arid conditions in the Qinling Mountains, China, a strong correlation between altitude and δ 2 H lipid (but not δ 13 C lipid ) highlights the potential to reconstruct and evaluate palaeoelevation and its interaction with local hydroclimate (Liu, 2021).These studies demonstrate both the complexity but also the valuable and detailed environmental issue which can be recovered using CSIA.
Where temperature and hydroclimate reconstructions are available from the same archive, the synchroneity or links between both larger and smaller scale climate drivers can be interrogated (e.g.Berke et al., 2014;Tierney et al., 2008;Muñoz et al., 2020;Stockhecke et al., 2021).In Lake Victoria, Africa, coherence between leaf wax δ 2 H hydroclimate and GDGT-inferred temperature records ('Quantifying amplitudes and rates of past temperature change') provided clear evidence for orbitally forced tropical climate since the Late Pleistocene, and highlighted the role of ENSO-related teleconnections in shaping climatic events such as the Younger Dryas (Figure 5) (Berke et al., 2012b).In Lake Elsinore (California), abrupt changes recorded by leaf wax δ 2 H in the Lateglacial (32-20 ka) were independent of GDGT-inferred temperature shifts and were attributed to changes in storm tracks (Feakins et al., 2019).In a marine sediment core offshore Sumatra, leaf wax δ 2 H challenged previous views of increased precipitation over the Indo-Pacific Warm Pool during the Last Glacial Maximum, which was attributed to regional differences in deglacial sea level and coastline configuration (Niedermeyer et al., 2014).In turn, new Holocene oscillations in the Indian Ocean precipitation could be linked to rainfall in East Africa via a 'precipitation dipole', rather than by ENSO (Niedermeyer et al., 2014).These examples are important for demonstrating that we can extend our understanding of the Lateglacial climate instability beyond ice and ocean dynamics, to include hydroclimate and atmospheric variability, especially in the low latitudes.

Reconstructing salinity using lake and marine sediments
Palaeosalinity reconstructions in the oceans and in lakes may provide an indication of changes in circulation (e.g. through changing water masses or currents) or hydroclimate (e.g.where enhanced freshwater inputs or increased evaporation can lead to lake salinity changes).In estuarine or coastal settings, salinity variations may also reflect changes in river discharge or the relative contribution of marine and freshwater as influenced by local changes in relative sea level.In this section we outline both biomarker distributions and CSIA which have detailed changes in salinity either in marine or lacustrine settings.

Sea-surface salinity as an indicator of circulation or sea-level changes
During the early U K 37 ′-SST calibration work ('Ocean and lake temperature reconstructions'), a potential salinity or polar water mass influence over the abundance of the haptophyte algae C 37:4 alkenone was determined (Rosell-Melé, 1998;Bendle et al., 2005), noting that this alkenone is not part of the U K 37 ′ index (Table 1).Subsequently, high C 37:4 values have been used to track expansion of (sub)polar water masses in the Atlantic, Pacific, and Southern Oceans across glacial-interglacial and million-year timescales (McClymont et al., 2008b;Martínez-Garcia et al., 2010).Elevated C 37:4 alkenone abundances (low salinity) have identified meltwater from Heinrich event icebergs reaching the Iberian Peninsula (Martrat et al., 2007), and glacial meltwater reaching the north-east Pacific (Sánchez-Montes et al., 2020).Although not specific salinity markers, terrestrially derived biomarkers in the iceberg-rafted debris (IRD)-rich Heinrich layers (Madureira et al., 1997;Rosell-Melé et al., 1997;van der Meer, 2007) confirmed the release of IRD and meltwater to the North Atlantic Ocean.Alternatively, large inputs of heavily altered carotenoids to southern Greenland, in the absence of IRD, suggested that an outburst flood occurred during the last interglacial (Nicholl et al., 2012).
More direct records of sea-surface salinity draw on the impact of changing salinity on D/H fractionation in seawater and during biosynthesis (e.g.Sauer et al., 2001;Englebrecht and Sachs, 2005;Schouten et al., 2006).Cultured haptophyte algae show that δ 2 H alkenone records salinity change (Engelbrecht and Sachs, 2005;Schouten et al., 2006), and may even be used to identify the source regions of alkenones transported to sediment drift sites (Englebrecht and Sachs, 2005).An early application in the eastern tropical Pacific used instrumental records to show that δ 2 H alkenone fluctuations recorded rainfall and river discharge in Columbia, and revealed reduced runoff during the last glacial compared to the Holocene (Pahnke et al., 2007).Combined δ 2 H alkenone and dinoflagellate cyst analysis showed substantial freshening of the Black Sea over the last ~3000 years, and refuted a hypothesis that salinity changes were responsible for changes to the haptophyte assemblage (van der Meer et al., 2008).In the southeast Atlantic, a decoupling of SST and salinity across multiple deglaciations has been recognized, whereby salinity (δ 2 H alkenone ) increased earlier than ocean warming (U K 37 ′ index); both changes pre-date the onset of deglaciation and may even play a role in triggering or facilitating ocean circulation change during glacial-interglacial transitions (Kasper et al., 2014;Petrick et al., 2015).In the Mediterranean Sea, δ 2 H alkenone confirmed a large drop in surface salinity at the onset of a Last Interglacial sapropel, supporting the hypothesis that these organic-rich layers were the result of precession-driven monsoon rains disrupting the circulation (van der Meer et al., 2007).As for leaf wax δ 2 H ('Reconstructing vegetation and hydrological change using compoundspecific stable isotope analysis'), care is needed to assess whether salinity change is the primary signal being recorded by sedimentary δ 2 H lipid , since it could also be impacted by factors including variations in growth rate (Wolhowe et al., 2009) and the algal species/genera (Schouten et al., 2006;van der Meer et al., 2008;Nelson and Sachs, 2014).
In coastal systems, salinity change can be a reflection of relative sea-level change.A fall in C 37:4 abundance (increased salinity) was used to identify relative sea-level rise in a Scottish isolation basin following the last deglaciation (Bendle et al., 2009).The relative contribution of mangrove species biomarkers (e.g.taraxerol) to intertidal sediments has also been explored as an alternative indicator of sea-level change (Versteegh et al., 2004;Koch et al., 2011), but local influences on sedimentation patterns and biomarker degradation require further investigation (He et al., 2018;Sefton, 2020).Both nalkane and taraxerol δ 2 H in mangrove systems show potential for isolating a biological response to changes in salinity (Ladd and Sachs, 2015).A salinity impact on mangrove water-use efficiency was also indicated by n-alkane δ 13 C in Australia (Ladd and Sachs, 2013).A challenge in low-latitude settings is to isolate a sea-level-driven salinity change from a hydroclimate impact on precipitation or seawater δ 2 H (e.g.Pahnke et al., 2007;Tamalavage et al., 2020).However, by combining pollen analysis with plant wax distributions and δ 2 H from a mangrove system in the Bahamas, the time-varying influences of changes in vegetation assemblage and precipitation could be disentangled for the Holocene (Tamalavage et al., 2020).Multi-proxy analyses thus show great potential for evaluating the relative influences of vegetation change, hydroclimate and sea-level-driven salinity variability in mangrove environments.

Lake salinity as an indicator of hydrological change
As observed in the marine environment (subsection above), high abundances of the haptophyte-algae C 37:4 alkenone have been recorded with low salinity in modern calibration studies of saline lakes (Liu et al., 2008(Liu et al., , 2011;;Song et al., 2016;He et al., 2020) and in comparisons between lake reconstructions and instrumental data (He et al., 2013).Qualitative palaeosalinity reconstructions using C 37:4 abundance in lake sediments have reconstructed late Holocene moisture fluctuations on the Northern Tibetan Plateau linked to solar irradiance (He et al., 2013), and identified the transition between marine and lake environments associated with ice-shelf expansion in north-east Greenland (Smith et al., 2023).However, not all lakes have recorded the C 37:4 alkenone (e.g.Toney et al., 2010), and seasonal biases in alkenone production may influence the reconstructions (He et al., 2020).Combined analysis of alkenone distributions and phylogenetic analysis in a suite of saline Chinese lakes (Yao et al., 2022) indicates that C 37:4 alkenone may reflect changing haptophyte groups rather than salinity, since the detected groups occupied different ecological niches.The presence of another salinity-sensitive indicator, the alkenone C 38:3Me , was detected during times of haptophyte assemblage changes consistent with fresher surface waters in a Pleistocene record from Lake Van, Turkey (Randlett et al., 2014).Palaeosalinity indices, such as the RIK 37 (ratio of isomeric ketones of C 37 chain length) index (Longo et al., 2016), capture salinity-driven shifts in haptophyte species composition and are reliable salinity proxies in oligohaline environments (Longo et al., 2016).
Salinity is also reflected in lake water δ 2 H and the biosynthesis of algal lipids: field calibration laboratory culture studies have demonstrated that the salinity is inversely related to the D/H fractionation of algal lipids (e.g.Sessions et al., 1999;Schouten et al., 2006;Sachse and Sachs, 2008;Schwab and Sachs, 2011;Ladd and Sachs, 2012;Nelson and Sachs, 2014;Englebrecht and Sachs, 2015;  and hydrological change using compound-specific stable isotope analysis' for discussions of other controls on δ 2 H lipids ).Mid-Holocene changes to the Indian Summer Monsoon have been detected using biomarker δ 2 H in a saline-alkaline lake in the core 'monsoon zone' of central India (Sarkar et al., 2015): more enriched δ 2 H in terrestrial leaf waxes and cyanobacteria, alongside increased abundance of the biomarker tetrahymanol (generated under saline conditions; Romero-Viana et al., 2012) reconstructed increased salinity and a lowering of lake levels after 6 cal ka BP (Sarkar et al. 2015).Archaeal GDGTs have also been used as palaeosalinity indicators based on ratios of archaeol, a biomarker for hypersaline archaea, and caldarchaeol, a cosmopolitan isoGDGT that is produced across a range of salinity conditions.The Archaeol and Caldarchaeol Ecometric (ACE) index (Turich and Freeman, 2011) has since been used as a qualitative lacustrine palaeosalinity proxy, showing that salinity increased due to a reduced water balance during periods of higher Lateglacial temperatures in southern California (Feakins et al., 2019).However, a study of 55 lakes in mid-latitude Asia has identified a threshold response in the ACE index, which suggests that it may only be effective in high lake salinity ranges (60 000-100 000 mg L −1 ) (He et al., 2020).

Reconstructing changes in sea ice extent
Early identification of elevated concentrations (>5-10%) of the abundant haptophyte algae C 37:4 alkenone in high-latitude marine samples suggested that low temperatures and/or low salinity in (sub)polar waters were important (see 'Sea-surface salinity as an indicator of circulation or sea-level changes').Subsequently, DNA analyses have demonstrated that high C 37:4 abundances can be more specifically linked to sea iceassociated haptophyte algae (Wang et al., 2021b).With further testing, this new evidence offers the potential for both sea ice and SST information to be simultaneously retrieved from alkenone data in the high latitudes.
Two related sea-ice biomarker proxies have been more extensively developed: specific highly branched isoprenoids (HBIs) usually synthesized in spring by particular iceassociated diatoms (see detailed review by Belt, 2018).In the Arctic, the mono-unsaturated alkene containing 25 carbon atoms is used ('IP 25 ', Belt et al., 2007) but this is not present in the Southern Ocean.Instead, the di-unsaturated HBI ('IPSO 25 ') is applied (Belt et al., 2016) (Table 3).Extensive evaluation of the HBIs, especially IP 25 , against diatom proxy data gives confidence in their ability to reconstruct sea-ice changes (Massé et al., 2008;Weckström et al., 2013).IPSO 25 is a relatively specific environmental indicator, reflecting the tendency for its producer, Berkeleya adeliensis, to live in platelet ice and the bottom layer of land-fast ice (Belt et al., 2016;Riaux-Gobin et al., 2000), and thus shows a strong signal of coastal production (Massé et al., 2011;Rontani et al., 2019).However, since HBIs have also been determined beyond the continental shelf edge, in the Scotia Sea (Collins et al., 2013), further investigation is required to fully evaluate the interpretation of IPSO 25 beyond the coastal regions.
A challenge for both HBI proxies is how to interpret the sea ice signal when IP 25 or IPSO 25 is absent.Absence could reflect compound degradation within the sea ice, water column or sediments (Belt, 2018), although recent work has confirmed IP 25 in pre-Quaternary sediments (Knies et al., 2014;Clotten et al., 2018).Alternatively, productivity by ice-dwelling diatoms may be minimal or absent under permanent sea ice cover if photosynthesis is restricted (Belt, 2018).To address the latter concern, the relative abundance of IP 25 or IPSO 25 can be compared with open-ocean productivity biomarkers (e.g.HBI III or brassicasterol for diatoms, dinosterol for dinoflagellates).Revised 'PIP 25 ' or 'PIPSO 25 ' indices have been proposed to describe this ratio (Table 3): an absence of both the sea-ice and open-ocean biomarkers yields a PIP(SO) 25 value of zero ('perennial sea ice'), whereas open-ocean only biomarkers yield a PIP(SO) 25 value of 1; values in between reflect seasonal sea ice presence (Belt & Müller, 2013).
IP 25 records have been important in assessing the role of sea ice in past climate changes.Relatively short historical sea ice records have been extended (Tesi et al., 2021).By filling in intervals of sparse historical data, abrupt changes in sea ice have been reconstructed for the last millennium (Massé et al., 2008).Millennial-scale fluctuations in spring sea-ice cover occurred to the north of Iceland during the Holocene and the last glacial-interglacial cycle (e.g.Müller et al., 2009;Hoff et al., 2016;Stein et al., 2017;Xiao et al., 2017;Sadatzki et al., 2020), including contrasting sea-ice conditions between the early/mid-and late Younger Dryas close to northern Norway (Cabedo-Sanz et al. 2013).Regional differences between the timing of expanded sea-ice cover were proposed to have contributed to millennial-scale variability in deepwater formation across the deglaciation (Figure 6) (Xiao et al., 2017).In the longer term, an increase in Bering Sea sea-ice cover and development of the seasonal advance and retreat of the sea ice margin occurred alongside the mid-Pleistocene transition ~1 Ma, which might have been important for influencing ice-sheet growth and increased deep ocean storage of carbon during glacial stages (Detlef et al., 2018).The transition from the warm Pliocene epoch into the Quaternary also saw an expansion of Arctic sea ice alongside the intensification of northern hemisphere glaciation ~2.7 Ma (Knies et al., 2014;Clotten et al., 2018).Although preservation over long timescales is promising, concerns have also been raised about the inherent instability of HBIs, meaning caution needs to be applied to interpretation of their presence/absence (Sinninghe Damsté et al., 2007).
IPSO 25 records have been integrated within several multiproxy studies.Expanded seasonal sea ice cover occurred during the last glacial stage in the Scotia Sea (Collins et al., 2013), and millennial-scale evolution of perennial and seasonal sea ice was recorded over the last deglaciation in the Amundsen Sea (Lamping et al., 2020).Multiple IPSO 25 records detail expansion and retreat of sea ice during the Holocene (Barbara et al., 2010(Barbara et al., , 2016;;Etourneau et al., 2013;Denis et al., 2010;Tesi et al., 2020;Ashley et al., 2021;Johnson et al., 2021).High-resolution analyses of the last ~400 years have shown that IPSO 25 can identify trends and cyclicity in seasonal and perennial sea ice cover, and links to ocean or atmospheric forcings (e.g.Campagne et al., 2015;Barbara et al., 2016;Vorrath et al., 2020).Differences in Holocene seaice histories between sites probably indicate the influence of local and regional circulation systems (Lamping et al., 2020;Vorrath et al., 2020), which are also expressed in the instrumental record (e.g.Parkinson, 2019).

Tracing biological productivity and biogeochemical cycling
Biomarker proxies implicitly record the flux of organic matter between different reservoirs of the Earth system.In this section, we outline biomarkers which have been used qualitatively to explore biogeochemical cycles in more detail by either detecting specific environmental conditions (e.g.biomarkers for methanogenic or methanotrophic micro-organisms) or for tracing changes in productivity and degradation (e.g.fluxes of biomarkers linked to specific producers).

Reconstructing biological productivity in lakes and the oceans
The source-specific nature of biomarkers allows for groups of producers to be traced in sedimentary systems, and to assess whether their productivity has changed in the past (Tables 2  and 3).When comparing the relative abundances of productivity markers, it is important to assess the potential impacts of bioturbation, remineralization and degradation of organic matter; these can be rapid and effective in oxic settings and could bias the target productivity signal (e.g.Leavitt, 1993;Arndt et al., 2013;Jessen et al., 2017).Intact pigments are particularly vulnerable to oxidation, UV radiation and associated processes of degradation, and usually have very low preservation in marine sequences (Reuss et al., 2005;McGowan, 2013).Better preservation may be recorded in lake sediments, but still more successfully with anoxic water columns, or with minimal sinking depths and benthic algae coverage (Leavitt, 1993;Hodgson et al., 2005;McGowan, 2013).
Pigment analysis has detected lake productivity oscillations in central Italy linked to warm-cold oscillations in the North Atlantic between ~15.0 and 28.0 cal ka BP (Chondrogianni et al., 2004), and changes in lake level linked to the onset of the African Humid Period in Ethiopia (Loakes et al., 2018).In East Antarctica, recolonization and succession of marine flora has been determined as the ice sheet and sea ice interacted through the Holocene (Hodgson et al., 2003).A distinctive pigment is isorenieratene (Table 3), a carotenoid pigment synthesized by green sulphur bacteria, making it a biomarker for a relatively uncommon but specific environment: photic zone euxinia (both anoxic and sulphidic) (Sinninghe Damsté et al., 2001).Isorenieratene has been instrumental in demonstrating that euxinic conditions developed during the Last Interglacial in the Mediterranean Sea associated with the formation of sapropels (Marino et al., 2007).Significantly, the co-recorded proxy data illustrated the role of increased runoff in altering Mediterranean circulation (see 'Sea-surface salinity as an indicator of circulation or sea-level changes') (Marino et al., 2007).
It is more common to find pigment degradation products in marine sediments, often alongside lipid biomarkers for other producers or degradation pathways (Table 3).Chlorophyll degradation products, chlorins ('Biological functions of biomarkers'), have been used to reconstruct export production, i.e. the organic matter which is removed from the surface ocean and stored longer term in the deep ocean or sediments (e.g.Petrick et al., 2018).Chlorins, alkenones, sterols and diols have reconstructed intensification or shifts in export production across multiple glacial-interglacial cycles linked to coastal upwelling systems (Petrick et al., 2018), highly productive oceanographic fronts (Cartagena-Sierra et al., 2021), sea-ice extent (Fahl and Stein, 1999) and changing nutrient inputs (e.g.Martínez-Garcia et al., 2011;Sánchez-Montes et al., 2022).In the Subantarctic Atlantic Ocean, a consistent pattern of elevated higher plant n-alkanes during glacial intervals aligned closely with dust peaks in Antarctic ice cores (Martínez-Garcia et al., 2009).In turn, lower SSTs and higher primary productivity (both reconstructed from alkenones) demonstrated close connections between ocean and atmosphere circulation, nutrient supply and potential glacial-stage CO 2 drawdown by the ocean through the Quaternary (Martínez-Garcia et al., 2011).A recent global-scale analysis of seafloor sediments flags the potential that alkenone concentrations may be dominated by primary productivity, and thus provide a potentially quantitative reconstruction of production over Quaternary timescales (Raja and Rosell-Melé, 2021).

Reconstructing sediment, organic matter and nutrient cycling
The presence of terrestrial biomarkers in marine sediments can enable an assessment of the links between ocean circulation and environmental change onshore as detailed above, but may also give insights into the transport pathways of terrestrial organic material and identify important connections between nutrient cycles and productivity alongside palaeohydrology.For example, flood events have been identified in estuarine sediments by increases to the C 31 /C 17 n-alkane ratio (Meyers, 2003), which were consistent with historical records of the Minjiang River, China, since the 1800s CE (Wang et al., 2014).A 'terrestrial to aquatic organic matter n-alkane ratio' (TAR, Table 3) has been used to record both dust and glacier-derived sediment inputs to the North Atlantic and Gulf of Alaska across multiple glacial-interglacial cycles (Naafs et al., 2012;Lang et al., 2014;Müller et al., 2018) with potential impacts on marine productivity (Müller et al., 2018;Sánchez-Montes et al., 2020).Biomarker fingerprinting of sediments eroded by the circum-Atlantic ice sheets has added to this detail, and determined the asynchroneity of IRD or meltwater release between different ice sheets (e.g.Stein et al., 2009;Rosell-Melé et al., 2011;Naafs et al., 2013;Hefter et al., 2017).
As well as tracing these land-ocean and land-lake transfers of organic matter, and describing or quantifying lake/ocean export productivity (subsection above), biomarkers can be used to trace biogeochemical cycling in two ways: (i) the presence of biomarkers generated under specific environmental conditions, e.g.anoxic settings; or (ii) the presence of diagenetic products of the original biosynthesized molecule, where the environmental controls on diagenesis are known.Although used to qualitatively describe organic matter formation, transport and reworking, there is emerging potential to consider biomarker concentrations or transformations as a way to quantify carbon burial and biogeochemical interactions including nutrient and oxygen availability.
In peatlands, biomarker tracers of biogeochemical cycling have been explored, due to the close links between peat water table depth, oxygen availability and the associated generation of greenhouse gases.For example, elevated concentrations of the anaerobic archaea-produced archaeol reflect rising water tables in peat sequences (Pancost et al., 2011) or enhanced methanogenesis during warm periods of the late Pleistocene and Holocene in Siberian permafrost (Bischoff et al., 2013).Methanogens are also probably the main source of isoGDGT-0 in peats (Basiliko et al., 2003, Pancost andSinninghe Damsté, 2003); by comparing iso-GDGT-0 and archaeol accumulation rates in a 16-kyr-old peat sequence from Hani, China, the long-term link between elevated levels of methanogenesis, high temperatures and high summer insolation was demonstrated (Zheng et al., 2019).
Biohopanoids are largely biomarkers of aerobic bacteria (Rohmer et al., 1992;Talbot et al., 2016), and include relatively simple C 30 hopanoids (e.g.diploptene), or more complex versions with additional side chains (bacteriohopanepolyols or BHPs; reviewed by Kusch and Rush, 2022).BHPs have a wide range of sources including methanotrophs, heterotrophs and phototrophs (reviewed by Talbot et al. 2016;Inglis et al., 2018;Kusch and Rush, 2022).Quaternary applications of BHPs in the Congo fan have demonstrated the correlation between elevated aerobic methane oxidation in the wetlands onshore and late Quaternary interglacial climates (Talbot et al., 2014) as well as a longer term shift ~1 Ma (Spencer-Jones et al., 2017).Variations in archaeol and diploptene δ 13 C values suggested links between the strength of the Asian monsoon and fluctuations in atmospheric methane concentrations (Zheng et al., 2014).Low δ 13 C diploptene have also traced the presence and small-scale spatial heterogeneity of methane-oxidizing bacteria (MOB), and therefore methane oxidation, in Alaskan thermokarst lakes (Davies et al., 2016).
Long-term insights into the nitrogen cycle have been developed using the bacteriohopanetetrol stereoisomer (BHTx), a tracer of anaerobic oxidation of ammonium (anammox) (Rush et al., 2014).For example, BHT-x demonstrated the link between higher temperatures and the intensification of oxygen deficiency zones in the Late Pleistocene in the Gulf of Alaska (Zindorf et al., 2020).Their study indicated that, unlike redoxsensitive trace metals, BHT-x is not impacted by dilution effects of high sedimentation rates.Ammonium oxidation has also been reconstructed using ratios of isoGDGT [2]/[3], indicating the presence of the archaea Thaumarchaeota: in the South China Sea, interglacials were shown to be characterized by concurrent increases in ammonium oxidation and δ 15 N-inferred N 2 fixation (Dong et al., 2019).
Transformation of the original biosynthesized compounds into recognizable products, under specific redox conditions, has also allowed changes in aerobic/anaerobic conditions to be traced in a range of environments.Interlinked changes to pH and water table explained the presence and down-core variations of an unusual hopanoid (the C 31 17α,21β(H)homohopane) in Holocene peats, which is usually only found in thermally mature organic matter (Pancost et al., 2003;McClymont et al., 2008a;Inglis et al., 2018).Transformation of sterols into stanols at the interface between oxic and anoxic conditions (Wakeham, 1989;Naafs et al., 2019)  conditions and water table depth (Naafs et al., 2019).In a marine sediment core, the different resistance to oxygenation of a plant wax n-alcohol and n-alkane was exploited to identify bottom current strength and thus duration of organic matter exposure to oxygenated waters across multiple millennial-scale and glacial-interglacial cycles (Martrat et al., 2007).
To assess the impacts of biogeochemical cycles on atmospheric CO 2 , the δ 13 C alkenone biomarker proxy showed early promise, drawing on the fractionation of stable carbon isotopes during haptophyte photosynthesis (Bidigare et al., 1997).However, recent work has demonstrated that CO 2 uptake by haptophytes is different at low CO 2 concentrations (Badger et al., 2019), indicating the need for careful interpretation of alkenone-based CO 2 reconstructions during the Quaternary.

Sedimentary records of humans and animals in Quaternary landscapes
Lipid biomarker analyses of sedimentary archives are increasingly used to characterize the presence, activities and impacts of humans and animals in the landscape, either as independent reconstructions or as complementary evidence in support of archaeological and palaeoecological anthropogenic reconstructions.Biomarkers also offer an alternative approach when levels of preservation are low or where archaeological excavation is not possible due to time, financial or logistical constraints (discussed in Brown et al., 2022).Biomarkers in archaeological remains contain a wealth of information about the origin of artefacts and deposits and their associated use (reviewed by Evershed, 2008); however, here we focus on sedimentary biomarker proxies that provide both direct and indirect evidence for the presence and environmental impacts of human and animals.For more information, we direct readers to the dedicated review of anthropic biomarkers in sediment archives (Dubois and Jacob, 2016).

Faecal biomarkers as direct sedimentary indicators of human and animals
Faecal steroid biomarkers (5β-stanols, bile acids), which are produced in the digestive tracts of mammals and deposited via excrement into the environment, present an opportunity to directly identify both animals and humans from sedimentary archives (reviewed by Bull et al., 2002).These compounds are well preserved within sedimentary archives over Holocene timescales (e.g. Simpson et al., 1998;D'Anjou et al., 2012;White et al., 2019;Schroeter et al., 2020;Brown et al., 2021).Different species produce different diagnostic distributions of faecal steroids due to differences in diets, digestive processes and gut bacteria (e.g.Leeming et al., 1996).Steroid ratios have therefore been used to distinguish between source organisms in investigations of modern faeces and archaeological deposits (e.g.Prost et al., 2017;Zocatelli et al., 2017;Shillito et al., 2020;Kemp et al., 2022), including through multivariate statistical analysis (Harrault et al., 2019).The presence of 5βstanols is not conclusive evidence of faecal deposition, since small amounts can be produced through the reduction of cholesterol in sedimentary environments (e.g.Gaskell and Eglinton, 1975;Bethell et al., 1994), but the application of sterol ratios and the tandem analysis of sterols and bile acids can be used to confirm faecal input and improve faecal source assignment (e.g.Prost et al., 2017).Identification of faecal sources is improved by characterizing steroid distributions of local reference dung to correct for within-species variability of sterol threshold values (Larson et al., 2022) and reference soils to account for in situ sterol transformation (e.g.Bull et al., 2002;Birk et al., 2012).
Interactions between seabirds and their environment have been particularly effective using faecal steroids as indicators of nearby colonies (reviewed by Duda et al., 2021).The changing impacts of penguin colonies and vegetation types (e.g.mosses vs lichens) on the West Antarctic Peninsula over the last 2400 years have been assessed from lake sediments (Wang et al., 2007).Local declines of northern common eider (Somateria mollissima borealis) populations in Arctic Canada and Greenland have been linked to changes in sea-ice concentrations during the Little Ice Age (Hargan et al., 2019), and Holocene little auk (Alle alle) population changes have been linked to the availability and stability of open waters (polynyas) in the sea ice (Ribeiro et al., 2021).
New insights into the presence and impacts of humans in past landscapes have occurred where faecal steroids have refined the timings of human arrival and settlement activities in locations such as northern Norway (D'Anjou et al., 2012), the North Atlantic Faroe Islands (Curtin et al., 2021), the Azores Archipelago (Raposeiro et al., 2021), the Pacific Cook Islands (Sear et al., 2020) and New Zealand (Argiriadis et al., 2018).Faecal steroids have also reconstructed the presence of humans and/or livestock (e.g.White et al., 2018;Vachula et al., 2019;McWethy et al., 2020;Elliott Arnold et al., 2021;Keenan et al., 2021;Ortiz et al., 2022), characterized longterm animal husbandry practices and land use (e.g.Mackay et al., 2020;Schroeter et al., 2020;Birk et al., 2021), and the diets of extinct species (e.g.van Geel et al., 2008;Sistiaga et al., 2014).Comprehensive modern characterization of east African megafauna also illustrates the potential for faecal sterol applications to inform conservation palaeobiology (Kemp et al., 2022).
Robust sedimentary faecal biomarker identifications of human presence in past landscapes are developed in combination with other sedimentary markers of anthropogenic activity such as pollen, charcoal, fire-derived lipid biomarkers (e.g.D' Anjou et al., 2012;Battistel et al., 2016;Section 8.2) and/or domesticated mammal sedaDNA (e.g. Brown et al., 2022), and are integrated with the existing historical and/or archaeologyical context.Current uncertainties associated with within-species variability of steroid distributions, contributions from environmentally transformed 5β-stanols, and steroid transportation, storage, secondary deposition and degradation processes (e.g.Birk et al., 2021;Keenan et al., 2021;Davies et al., 2022;Lawson et al., 2022) present a range of opportunities for further analysis to refine steroid identification of faecal sources and enhance their applications as anthropogenic and mammalian tracers in Quaternary science.

Biomarkers of burning and agricultural activity as indirect indicators of human activity
Pyrogenic biomarkers can enhance understanding of fire histories since their signatures and concentrations record information on the fuel type and conditions during the fire such as burn intensity and moisture content, as demonstrated through modern burning experiments (e.g.Oros and Simoneit, 2001;Karp et al., 2020) and palaeo-comparisons with macro-and micro-charcoal (e.g.Elias et al., 2001;Schreuder et al., 2019a).
PAHs are produced during the incomplete combustion of biomass (reviewed by Richter and Howard (2000) and Lima et al. (2005)).PAH compound distributions represent combustion conditions, vegetation fuel type and transport pathways (Karp et al., 2020) and can be used to distinguish between local and regional burning events (e.g.Vachula et al., 2022).Many PAHs can be atmospherically transported across thousands of kilometres, although some compounds, such as benzo[a]pyrene, have lower modelled half-life transport distances of ca.500 km (Halsall et al., 2001).PAHs are produced by a wide range of burn temperatures (ca.200-700 °C; Lu et al., 2009), but higher concentrations are produced under high-intensity burning temperatures of 400-500 °C and during the combustion of woody rather than grassy vegetation (Karp et al., 2020).Palaeo-PAH records may therefore be biased towards wildfires and sensitive to changes in fuel type and/or fire regime.Whilst PAHs can be released from petrogenic sources (e.g.Wakeham et al., 1980), pyrogenic inputs can be identified using relative distributions of PAHs (e.g.Stogiannidis and Laane, 2015) or through comparisons with other fire proxies (e.g.Ruan et al., 2020;Tan et al., 2020).Long-term records of PAH fire histories have tracked human settlement and activity in the late Holocene in northern Norway (D'Anjou et al., 2012), East Africa (Battistel et al., 2016) and New Zealand (Argiriadis et al., 2018) and characterized the advent of hominin pyrotechnology in the Middle Palaeolithic (Brittingham et al., 2019).PAHs from lake sediments have also tracked industrial emissions such as combustion of coal (e.g.Meyers, 2003) and other fossil fuels (e.g.Guo et al., 2022); anthropogenic pollution contributions must be considered if using PAHs to reconstruct fire histories over the industrial period.
Levoglucosan and its isomers (mannosan and galactosan) are monosaccharide anhydride (MA) compounds that are specific palaeo-fire proxies (reviewed by Simoneit, 2002 andBhattarai et al., 2019) since they are exclusively formed during the combustion of cellulose (Simoneit et al., 1999) at burn temperatures of ca.150-350 °C (e.g.Kuo et al., 2008).MAs can travel hundreds to thousands of kilometres transported by wind and rivers (e.g.Mochida et al., 2010;Zennaro et al., 2014).Ratios of levoglucosan, mannosan and galactosan can reveal the type of biomass involved in burning events (e.g.Fabbri et al., 2009;Kirchgeorg et al., 2014) and combustion conditions (e.g.Kuo et al., 2011).Lake sediment comparisons of macroscopic charcoal and MAs from the Mayan Lowlands, Guatemala, demonstrated the advances of combining these fire proxies to enhance understanding of palaeo-fire regimes at different spatial scales (Schüpbach et al., 2015).Offshore levoglucosan records have confirmed vegetation changes associated with the late Quaternary megafaunal extinction in south-eastern Australia (Lopes dos Santos et al., 2013b) and demonstrated increased burning linked to vegetation change and human settlement in sub-Saharan north-west Africa 60-50 ka (Schreuder et al., 2019b).MA records from ice cores have been successfully applied to track post-Last Glacial Maximum and Holocene fire intensity and burning type at regional to semi-hemispheric scales (e.g.Zennaro et al., 2014;Battistel et al., 2018;Segato et al., 2021;Chen et al., 2022).Combustion-derived derivatives of lignin phenols, monosaccharide molecules and diterpenoids are also major components of smoke particulate matter and can be detected in sediment archives (Oros and Simoneit, 2001).
Evidence of crop cultivation and processing can characterize the timings of human presence and the types of activities taking place in past landscapes.Although not every cultivar has known specific lipid biomarkers, millacin is a marker of the introduced broomcorn millet in well-defined botanical settings (e.g.Jacob et al., 2008aJacob et al., , 2008b;;Bossard et al., 2013).Fluxes of millacin detected in lake sediments have, for example, traced the introduction, intensification and failure of millet cultivation since the Bronze Age in the French Alps, and comparisons with contemporary palaeohydrological reconstructions have demonstrated climatically driven downturns in millet cultivation in the Hallstatt period (Jacob et al., 2008a).Other cultivar biomarkers include cannabinol, a marker of hemp that can be used to identify processing activities (retting) from sediment archives (e.g.Lavrieux et al., 2013;Schmidt et al., 2020;Rull et al., 2022), and palmitone, a marker of Colocasia esculenta Schott (taro) (e.g.Krentcher et al., 2019).

Conclusions and future outlook
Biomarkers have emerged as valuable parts of the Quaternary science toolkit, due to both quantitative and qualitative insights into past environmental changes, and because multiple biomarkers (and thus multiple environmental signals) can be recovered from single samples.Analytical developments and improved understanding of the processes underpinning the wide range of biomarker proxies outlined here have also led to data that have been both novel and complementary to more established Quaternary science approaches.
The major impacts of biomarker analyses have so far come from the quantification of temperature changes, and detailed assessments of the interactions between vegetation change and hydroclimate.The results are important in spanning a wide range of timescales, from annual/decadal through to the longterm evolution of Quaternary climates at glacial-interglacial and longer timescales.In considering future climate projections, both the quantitative and the qualitative insights gained from biomarker reconstructions have enabled data-model comparison and data-model assimilation to be undertaken across a wide range of timescales, including the pre-Quaternary (Tierney et al., 2020;Masson-Delmotte et al., 2021).In addition to providing valuable palaeoclimatic insights, biomarkers are increasingly being used to directly identify human impacts on the environment both pre-dating and through the Industrial era, thereby providing essential long-term context to advance our understanding of the resilience of ecosystems and societies.
Continued efforts to better constrain quantitative calibrations of temperature, salinity, sea ice and precipitation will further enhance our biomarker reconstructions.Communitywide collaborations have been important for advancing our understanding and application of palaeoenvironmental proxies and their uncertainties (e.g.Schouten et al., 2013 for TEX 86 ;Belt et al., 2014 for IP 25 ); similar approaches could assist with advancing our understanding of more recently developed or more qualitative biomarker proxies (e.g.anthropogenic markers).With the increasing application of (seda)DNA approaches to identify and understand the biomarker producers (e.g.Wang et al., 2019b;Theroux et al., 2020), more nuanced interpretations of past temperature or other environmental changes are also likely to result from reduced uncertainty estimates and through advances in our understanding of signals related to key producers and their potentially varied responses to factors including seasonality and nutrient availability.There is therefore the potential to add to the rich environmental information provided by both biomarkers and other geochemical and palaeoecological proxies, with new assessments of biogeochemical cycling, sea ice evolution and human-environment interactions, as well as new data on how that organic matter has been preserved, recycled and transported through palaeoenvironments.
In this review, we have outlined some of the many, diverse ways in which biomarkers have advanced understanding of Quaternary environments.The biomarker toolkit is continually evolving, aided by advances in instrument capabilities which are presenting new opportunities to analyse smaller sample sizes and a greater diversity of Quaternary archives.For example, improvements in detection limits facilitated by high-resolution mass spectrometry present opportunities to expand the suite of palaeoenvironmental proxies that can be analysed from a single sample, and extend applications where sample sizes are limited and/or biomarker concentrations may be low (e.g.varved sediments, ice cores and/or highly resolved sedimentary records).In turn, untargeted analysis of environmental mass spectrometry spectral data, such as hierarchical clustering (e.g.Bale et al., 2021) and the application of information theory and molecular networking (e.g.Ding et al., 2021), yields highly detailed molecular information, with the potential to provide unprecedented levels of information about environmental contributions as well the identification of as yet unknown biomarkers, which may prove to be of ecological and environmental significance.In addition, there is great potential to expand compound-specific analyses, which have already yielded detailed insights into past hydroclimate, productivity and CO 2 , by extending the range of biomarkers that can be analysed.A rapidly advancing area of biomarker research is radiocarbon analysis of individual lipids, or groups of lipids, which has already demonstrated that different pools of organic matter are being (re)worked and transported through river systems today (e.g.Galy & Eglinton, 2011;Eglinton et al., 2021;Feng et al., 2013) and have been in the past (Bliedtner et al., 2020).Biomarker radiocarbon analysis shows great potential to not only enhance our understandings of Quaternary sedimentary environments and processes, but also to improve chronological controls through compound-specific radiocarbon analysis.Biomarkers have therefore made a wealth of contributions to Quaternary science, and the continued advances in this field of research offer many opportunities to extend our understandings of Earth systems in the past, present and future.
Patient consent statement.Not applicable.
Permission to reproduce material from other sources.We have applied to the publisher for permission to present the maps we adapted in Fig. 5.

Figure 1 .
Figure 1.Using biomarkers to trace a wide range of environmental processes.Examples include biomarkers of climate change (e.g.temperature, precipitation, sea ice), ecosystem change (e.g.vegetation cover, productivity dynamics and fire regimes), biogeochemical cycling (e.g.methane production), sediment transport (e.g.soil residence time and land-ocean interactions), and human-environment interactions (e.g.presence of humans and animals and agricultural activity).Biomarkers can be transported between terrestrial ecosystems and to marine environments by rivers, surface water run-off, erosional processes, wind and melting ice.Abbreviations: IP 25 (Ice Proxy with 25 carbon atoms), IPSO 25 (Ice Proxy Southern Ocean with 25 carbon atoms), isoGDGT (isoprenoidal glycerol dialkyl tetraether), brGDGT (branched glycerol dialkyl tetraether), BHPs (bacteriohopanepolyols). [Color figure can be viewed at wileyonlinelibrary.com]

Figure 2 .
Figure 2. Two examples of biomarker distributions containing mixtures of aquatic and vascular plants.Analysis is by chromatography, whereby individual compounds are separated according to their size and chemical structures.The sample is injected at time zero, and the size of the peak corresponds to the abundance of that molecule in the sample.(a) Gas chromatogram of the apolar compounds recovered from a lake or marine sediment sample, showing a mixture of aquatic and vascular plant inputs; (b) gas chromatogram of the polar compounds recovered from a peatland sample, showing a mixture of vascular plant inputs.

Figure 4 .
Figure3.A selection of palaeotemperature biomarkers, detailing the different chemical properties that can be used to identify specific markers and their relationships to biological and environmental variables.(a) TEX 86 (TetraEther indeX of tetraethers consisting of 86 carbon atoms) temperature proxy is calculated using the relative distributions of isoGDGTs (iso-GDGT-1, iso-GDGT-2 and iso-GDGT-3) and the crenarchaeol regioisomer(Schouten et al., 2002).Blue circles highlight the number of cyclopentane moieties, and the yellow circle highlights the presence of a cyclohexane ring.(b) MBT′ 5Me (Methylation of Branched Tetraethers using the 5-methyl isomers) temperature proxy in soils is calculated using the relative distributions of 5-methyl brGDGT(de Jonge et al., 2014).Blue circles highlight the presence and number of cyclopentane moieties and green circles highlight the presence and number of methyl groups in the α and/or ω-5 position.(c) U K 37 ′ temperature proxy in freshwater and marine environments is calculated using the relative distributions of the di-and tri-unsaturated alkenone distributions(Prahl and Wakeham, 1987).The chain lengths of the two alkenones are the same (C 37 = 37 carbon atoms), but the number of double bonds increases from two to three (highlighted by red circles).(d) LDI (long chain diol index) temperature proxy in freshwater and marine environments is calculated using the relative distributions of C 28 and C 30 1,13-and C 30 1,15-alykl diol distributions(Rampen et al., 2012; 2014).Compounds vary in terms of chain lengths (C 28 = 28 carbons atoms and C 30 = 30 carbons atoms) and the location of the midchain alcohol group (C 13 or C 15 ; highlighted by the orange circles).(e) RAN 15 temperature proxy in soils is calculated using the ratio of anteiso to normal 3-hydroxy C 15 fatty acid(Wang et al., 2021a).Green circle highlights the methyl-substituent located on the antepenultimate carbon atom.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 5 .
Figure 5. Terrestrial and marine biomarker reconstructions of environmental change in east Africa since the late Pleistocene.(a,b) Palaeoclimate reconstructions from Lake Victoria (Berke et al., 2012): (a) TEX 86 palaeotemperatures and (b) palaeoprecipitation record from ice volume-corrected δ 2 H of the C 28 leaf wax fatty acid methyl ester (FAME) with error bars (grey lines) representing the mean error of replicated analyses for each sample.(c,d) Palaeoclimate reconstructions from Lake Tanganyika (Tierney et al., 2008); (c) TEX 86 palaeotemperatures and (d) palaeoprecipitation record from ice volume-corrected δ 2 H of the C 28 leaf

Figure 6 .
Figure 6.Schematic illustration of biomarker-(IP 25 ) inferred changes in spring/summer sea ice extent (white shadings) between (a) the Bølling/ Allerød, (b) Younger Dryas and (c) Early Holocene (adapted from Xiao et al., 2017; see the original figure for the detailed map key).Atlantic Water advection is represented by red arrows and cold Polar waters from the Arctic Ocean are represented by blue arrows.(d,e) Examples of the IP 25 records used to develop the sea ice maps in a-c.(d) Most northerly IP 25 record of sea ice presence (Yermak Plateau, denoted in purple; Müller et al., 2009) and (e) most southerly IP 25 record of sea ice presence (north of Iceland, denoted in green; Xiao et al., 2017) included in the schematic maps.[Color figure can be viewed at wileyonlinelibrary.com] and soils (~4.8 °C, De Jonge et al., 2014;

Table 3 .
Overview of the key proxies for palaeo-productivity and biogeochemical cycling and sediment transport.
c represent a constant (fixed number) in the equation, as standard mathematical nomenclature.