Introduction

Lipid biomarkers are ubiquitously distributed and excellently preserved in soils and aquatic sediment, and have been regarded as a suitable proxy for paleoclimate and paleoecological studies at different time scales (Castañeda and Schouten 2011; Huang and Meyers 2019; Summons et al. 2021; Chen et al. 2022). Among these compounds, fecal sterols are important biomarkers for mammal excrement and have been widely used to indicate sewage contamination in industrial and population centers over the past several decades (Schönning et al. 2002; Martins et al. 2018; Cabral et al. 2019). With increasing scientific expeditions and tourism in polar regions, fecal sterols are also used to assess possible impacts of human activities on regional environments (Venkatesan and Mirsadeghi 1992; Martins et al. 2002, 2014; Leeming et al. 2015; Hargan et al. 2018) as well as the migration history of early human and seabirds (D'Anjou et al. 2012; Cheng et al. 2016; Vachula et al. 2019).

Mammals and birds exhibit different distribution characterizations of fecal sterols (Bull et al. 2002). For example, cholesterol comprises over 98% of total sterols in fresh penguin guano (Martins et al. 2002; Leeming et al. 2015) and generally is the most abundant sterol in lacustrine sediment influenced by animal excrement (Ceschim et al. 2016). Cholestanol is regarded as the product of microbial hydrogenation of cholesterol by aerobic bacteria (Bull et al. 2002). The variation of cholesterol and cholestanol contents in ornithogenic sediment profiles has been successfully used to reconstruct historical changes of penguin populations in Antarctica (Wang et al. 2007; Huang et al. 2010; Hu et al. 2013; Zheng et al. 2022). Coprostanol and epi-coprostanol are predominantly distributed in human waste accounting for 50% of total sterols, while campestanol and stigmasterol are high in mammal excrement (e.g., cows and sheep; D'Anjou et al., 2012). Therefore, the ratio of coprostanol/stigmasterol is deemed as an indicator for activities of humans and animals in polar region (D'Anjou et al. 2012; Vachula et al. 2019). In addition, fecal sterols (e.g., coprostanol) were also present in marine mammal (e.g., whales, seals; Leeming et al. 2015) feces and have been used as a proxy for delineating historical seal population changes in Antarctica (Huang et al. 2011). Considering the importance of fecal sterols in evaluating human activities and paleoecological studies in polar regions, it is particularly important to accurately identify their sources and biogeochemical transformation processes in sediment.

According to a previous study, there are three possible sources for fecal sterols (coprostanol) in Antarctic aquatic sediment including human waste, marine mammal excrement, and microbial hydrogenation of sterols by bacteria under anoxic sedimentary conditions (Leeming et al. 2015). For example, human activities such as the discharge of sewage are the predominant source of fecal sterols in the vicinity of the research stations in Antarctica (Montone et al. 2010; Martins et al. 2014, 2018). Besides human activities, marine mammals (e.g., seals) also can produce significant quantities of coprostanol (Venkatesan and Santiago 1989; Leeming et al. 2015). Beside the two above-mentioned sources, in situ microbial hydrogenation of sterols by bacteria under anoxic conditions can contribute significantly to the coprostanol in aquatic sediment. As reviewed by Kenny et al. (2020), cholesterol can be converted to coprostanol by human gut microbiota under anoxic condition. For example, many anaerobic bacteria (genus Eubacterium, Bifidobacterium, Clostridium, and Lactobacillus) can convert cholesterol to coprostanol (Gérard et al. 2007; Kriaa et al. 2019; Kenny et al. 2020; Juste and Gérard 2021). Actually, the conversion of cholesterol to coprostanol through the hydrogenation process was observed in anoxic conditions (Gaskell and Eglinton 1975, 1976; Nishimura 1982; Jen and Han 1996; Speranza et al. 2018). For example, high concentration of coprostanol was found in anaerobic sludge digestion due to the diagenetic transformation of cholesterol (Ahmad and Eskicioglu 2019). Nonetheless, no research combined with sequencing data and redox proxies has been conducted to confirm whether cholesterol can be converted to coprostanol by microorganisms in Antarctic lacustrine sediment. Furthermore, the factors that control the conversion of cholesterol to coprostanol remains elusive.

To answer the questions on whether such conversion can occur and how in situ sediment conditions determine the conversion of cholesterol, we chose four sediment profiles (AC2, BI, CH1, and IIL1) collected at North Victoria Land, Ross Sea, Antarctica. We explored the fecal sterols in these sediment profiles, and analyzed different environmental samples including fresh penguin guano and seal excrement, ornithogenic soils, microbial mats, terrestrial moss and soils for comparative purposes. We also compared fecal sterol distributions with the redox proxy such as Oxygen Index, active iron Fe(III)/Fe(II) and Fe/Mn ratios and bacterial diversity obtained by the high-throughput sequencing of 16S rRNA gene. This study aimed to accurately identify the source of fecal sterols in Antarctic lacustrine sediment including human and marine mammal wastes and microbial transformation, and to evaluate whether the conversion of cholesterol to coprostanol by bacteria occurs under anoxic conditions. Our study helps understand the biochemical transformation of fecal sterols in Antarctic lacustrine sediment, and lays the foundation for further reconstructing paleo-environmental and paleo-ecological changes in Antarctica.

Material and methods

Setting and sampling

The Ross Sea is a deep embayment in Antarctica with an average depth of 500 m. The weather in this region is extremely severe due to the presence of three different air masses from the Ross Sea, the Ross Ice Shelf, and Victoria Land (Monaghan et al. 2005). The annual mean air temperature is -18 °C, and the temperature can reach -50 °C and 8 °C in austral winter and summer time, respectively (Nie et al. 2016).

All these four sediment cores (AC2, BI, CH1 and IIL1) were collected using PVC tubes (about 12 cm diameter) that were hammered into the sediment layer. After transport back to the laboratory, all these sediment cores were sectioned at 0.5 cm. AC2 sediment profile was sampled from a small pond at Adélie Cove (74°46′05.00″S, 164°00′57.7″E), Ross Sea in February 2016 (Fig. 1; Supplementary Table S1). There is no research station in this region. More than 11,438 Adélie penguin breeding pairs inhabit Adélie Cove (Humphries et al. 2017). The detailed description of the sampling site has been reported by Povero et al. (2002). Specifically, Adélie Cove is a small deep V-shaped bay located in coastal Terra Nova Bay. The climate is extremely severe because of strong katabatic winds with low air temperature and precipitation. Specifically, the core (16.5 cm in length) was generally dark gray with a foul smell, and penguin eggshells and feathers were found throughout the sediment profile.

Fig. 1
figure 1

Study area and sampling site on the North Victoria Land, western Ross Sea, Antarctica. (a) Ross Sea. (b) Beaufort Island. (c) Inexpressible Island

Full size image

The Beaufort Island (BI) sediment core (166°58′23.6″E, 76°58′23.6″S) was collected from a small catchment near an active Adélie penguin colony in January 2010 (Fig. 1). Beaufort Island (166°58′ E, 76°58′ S) is 5 km long and about 2.5 km wide. Most of the island is covered by snow and ice and the maximum elevation is 771 m above sea level (masl). No research stations exist on this island. More than 60,000 breeding pairs of Adélie penguins inhabit Beaufort Island, most on the southwest end, with a small Emperor penguin colony at the north end as well (Seppelt et al. 1999; Lynch and LaRue 2014). There are some small lakes or ponds near the Adélie penguin colony at the southwest end of the island that receive input from penguin guano. Our previous studies indicated that the source of organic matter in the BI sediment profile mainly derived from penguin guano and aquatic algae (Chen et al. 2013; Liu et al. 2013; Nie et al. 2014; Chen et al. 2020). The detailed description of the sampling site and sectioning of the sediment profile used in this study have been described by Nie et al. (2012). Briefly, the core BI (20.5 cm in length) gradually transforms from brown clay into black fine-grained ornithogenic soils from the bottom to top. The chronology of the BI sediment profile was determined by AMS14C and 210Pb-137Cs dating, with the results reported in our previous study (Chen et al. 2015), and the inferred age of bottom section from the BI profile was ~ 1640 AD.

CH1 sediment core was collected from a small pond located at Cape Hallett (72°19′S; 170°16′E), Ross Sea in January 2016 (Fig. 1). Cape Hallett peninsula is about 32 km long and 8 km wide including four parts based on landscape including Seabee Hook, Flat, Summit and Scree slope (Green et al. 2015). Seabee Hook is almost completely occupied by an Adélie penguin colony and a large number of abandoned penguin colonies can be found in the Flat area. More than 60,000 breeding pairs occupy on the Seabee Hook (Gordon 2003). There formerly was a research station built by American and New Zealand personnel in 1968, but it was abandoned in 2006. The climate is extremely cold with the highest monthly average air temperatures at -1.35 °C (January). Core CH1 (15 cm in length) is also largely influenced by penguin guano with a generally dark gray color and foul smell. Abundant penguin remains (eggshells, feathers and bones) can be found in the sediment profile. Additionally, core CH1 is near (about 500 m) the abandoned research station.

The IIL1 sediment core was collected from small pond on Inexpressible Island, Terra Nova Bay in January 2016. A detailed description of the sampling site is provided by Wei et al. (2016) and Chen et al. (2019). Inexpressible Island has an area of approximately 50 km2 (Fig. 1). A large number of abandoned Adélie penguin colonies are located there (Baroni and Orombelli 1994; Emslie and Patterson 2007), and an active colony of more than 20,000 breeding pairs occupy ice-free terrain surrounding Seaview Bay (He et al. 2017). The detailed sectioning of the IIL1 sediment profile has been reported by Wei et al. (2016). Core IIL1 (54 cm in length) is largely influenced by penguin guano with a dark color and foul smell below 26 cm. From 26 to 20 cm depth, the influence of penguin guano gradually decreased. The top 20 cm of the IIL1 sediment core is mainly microbial mat deposition with a gray-yellow color. Eggshells, bones and feathers can be found in the sediment profile. The age-depth model of the IIL1 sediment profile has been published in a previous study (Jin et al. 2021), and the bottom (oldest) age is dated to ~ 1659 year BP.

For comparison, two fresh Adélie penguin guano samples were collected from Cape Crozier, Ross Island (Table S1); two fresh seal excrements were collected from East Antarctica (Zhongshan station) and the Antarctic Peninsula (Great Wall Station, King George Island); three ornithogenic soils and two soils under moss samples were sampled at North Victoria Land. Terrestrial moss samples were sampled from Fildes Peninsula (King George Island, Antarctic Peninsula) and Campo Icarus, and Cape Irizar in the Ross Sea (Table S1). All samples were kept at - 20 °C prior to analysis.

Analysis of physicochemical properties

The Oxygen Index, Fe(III)/Fe(II), and Fe/Mn data are cited from our previous studies (Lou et al. 2016; Zheng et al. 2022). Oxygen Index represents the oxygen content of the organic matter, and can be used to reflect the redox nature of sediment (Meyers and Lallier-Vergès, 1999; Jacob et al. 2004). The OI value is calculated by the sum of CO (S3CO) and CO2 (S3CO2) released during pyrolysis, normalized to total organic carbon (TOC). The TOC was determined by potassium dichromate oxidation-capacity (Liu et al. 2013). Rock–Eval analysis followed the procedure described in Sanei and Goodarzi (2006), and was measured at the State Key Laboratory of Organic Geochemistry, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (RockEval-6 Standard, Vinci Technologies). Active iron is extremely sensitive to the variability of sedimentary environment and Fe(III)/Fe(II) values reflect the redox conditions in aquatic sediment (Haese et al. 1997). The analysis of active iron in the BI sediment profile was followed by the method of Haese et al. (1997). The Fe/Mn ratios are good indicator of redox conditions in lacustrine sediment with a higher value indicating a reducing environment (García-Rodríguez et al. 2021). The composition of Fe and Mn was measured by inductively coupled plasma optical emission spectrometry (ICP-OES) after microwave digestion (Perkin Elmer 2100DV; Zheng et al. 2022).

The contents of total organic matter in the BI and IIL1 sediment profiles are cited from Liu et al. (2013) and Wei et al. (2016), respectively. The δ13C values of total organic matter (δ13CTOC) in the BI and IIL1 sediment profiles were cited from our previous studies (Liu et al. 2013; Wei et al. 2016; Jin et al. 2021). The methods of carbon stable isotope analysis using the sealed tube combustion method were described in detail by Wei et al (2016) and Liu et al. (2013).

Lipid extraction and quantification

Organic geochemical analysis of AC2 and BI sediment cores and environmental samples (seal excrement, penguin guano, microbial mats, soils, terrestrial moss, and ornithogenic soils) in this study followed the general procedure described in Vachula et al. (2019). Briefly, the freeze-dried samples (~ 1 g) were homogenized and extracted with dichloromethane (DCM): methanol (9:1 by volume) using an accelerated solvent extractor (Dionex ASE 200) at Brown University, Providence, RI. The total lipid extract was dried under N2 and separated into acid (4% acetic acid in ether) and neutral (DCM: isopropanol) fractions using aminopropylsilyl gel columns. The neutral fraction was further separated into N1 (Hexane), N2 (DCM), N3 (Hexane: Ethyl acetate; 3:1 by volume), N4 (MeOH) by flash column chromatography using silica gel (40–63 μm, 60 Å). The N3 fraction was silylated with N,O-bis(trimethylsilyl) trifluoroacetamide (BSTFA) prior to quantification.

Analytical methods for CH1 and IIL1 sediment followed the general procedure from Cheng et al. (2016). Briefly, about 0.1 g freeze-dried sediment samples were sonicated with 10 mL dichloromethane (DCM) for 10 min with activated copper, repeated three times. The total lipid extract was concentrated to ~ 0.5 mL under N2 at room temperature and was cleaned using LC-Si SPE column eluted with 15 mL DCM. The eluted fraction including sterols and stanols was dried under a gentle flow of N2, CH1 and IIL1 sediment samples were analyzed using high temperature Gas Chromatography & Q Exactive Orbitrap Mass Sepctrometer.

Previous studies have compared the extraction efficiency between ASE and ultrasonication (Heemken et al. 1997), which showed that the recoveries of these two methods are equal; thus, our methods do not affect our interpretation in this study. The samples were quantified using an Agilent 6890N GC-FID system with an Agilent DB-1 column (30 m × 320 μm × 0.17 μm). The GC test method followed Longo et al. (2013). The initial temperature was 50 °C, followed by heating at 20 °C min−1 to 255 °C, then heating at 3 °C min−1 to 300 °C, and finally heating at 10 °C min−1 to 320 °C where it was held for 15 min. The lipid concentrations of the samples were determined using the peak areas obtained from the GC-FID results. The same GC column and oven temperature program were used for GC–MS analyses of selected samples for compound identification, which were performed on an Agilent 6890 GC interfaced to a 5973N quadruple mass spectrometer. Compounds were identified by comparison of GC retention times and mass spectra with published literature (Vachula et al. 2019). Compounds were quantified using an internal standard 5α-cholestane and by the proportional relation of the peak areas for each biomarker (supplementary materials, Tables S2–S6 and Fig. S1). The concentration of cholesterol and cholestanol in BI sediment profile was published by Zheng et al. (2022).

DNA extraction, PCR, sequencing and data analysis

DNA was extracted from 0.5 g of frozen sediment (AC2, CH1, and IIL1) using a FastDNA SPIN Kit for Soil (MP Biomedicals, Solon, OH, USA) following the standard protocols as previously described (Zhao et al. 2018; Chen et al. 2021). BI sediment profile was not analyzed due to lack of material. The concentration of each DNA extract was checked using a NanoDrop ND-2000 spectrophotometer (NanoDrop Technologies Inc., Wilmington, DE, USA). The 16S rRNA gene (V4 region) was amplified by primers 515F/806R using the Illumina MiSeq platforms (Bates et al. 2011). Target PCR products were then sequenced on an Illumina HiSeq 2500 sequencer (PE 250) at the Magigene Biotechnology Co Ltd (Guangzhou, China).

Sequencing data were processed using the software QIIME2 (version: 2021.2) (Bolyen et al. 2019). Specifically, sequences were denoised and clustered to amplicon sequence variants (ASVs) by DADA2 (Callahan et al. 2016), the dada2 denoise-single method with the parameters –p-trim-left 56, –p-trunc-len 315 was applied during the analysis. The generated ASVs were assigned taxonomy by the classify-sklearn method of q2-feature-classifier plugin (Bokulich et al. 2018) against SILVA reference alignment release 138 (Quast et al. 2013). The ASVs identified to be the "Eubacterium_coprostanoligenes_group" in the Silva database were selected to build the phylogenetic tree, and the full length 16S rRNA references sequences were downloaded from Silva database (https://www.arb-silva.de/). The sequences were aligned using MAFFT (Katoh and Standley 2013) and then trimmed by trimAL (Capella-Gutiérrez et al. 2009) with -automated1 parameter. The phylogenetic tree was built using IQ-TREE 2 (Minh et al. 2020) with the model MFP, 1000 ultrafast bootstraps and visualized by tvBOT (Xie et al. 2023).

Results

Fecal sterols distribution in environmental samples

The abundance of cholesterol in fresh seal excrement and penguin guano was very high, with the ratio of cholesterol/total sterols over 98% (Table 1). Most of seal excrement contained less than 1% coprostanol, and coprostanol and epi-coprostanol were not detected in penguin guano (Table 1), while cholesterol accounted for over 90% of the total sterols. Cholesterol was also predominant in microbial mats, soils, terrestrial moss, and ornithogenic soils sampled from modern penguin colony, with a relative abundance over 95% (Table 1), but coprostanol was not detected in all these environmental samples.

Table 1 The average abundance of cholesterol, cholestanol, coprostanol and epi-coprostanol in total C27 sterols in Antarctic environmental samples
Full size table

Fecal sterols distribution in lacustrine sediment profiles

Four C27 fecal sterols (cholesterol, cholestanol, coprostanol, and epi-coprostanol), were detected in the AC2, BI, CH1, and IIL1 sediment profiles (Table 2). The most abundant sterols were cholesterol in the AC2, BI, and IIL1 sediments, but coprostanol for CH1 sediment samples (Supplementary Figs. S2 and S3). Epi-coprostanol was not detected in the CH1 and IIL1 sediment.

Table 2 The average abundance of fecal sterols and ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in these four sediment profiles
Full size table

The ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) can be used as organic proxies to trace the sources of fecal sterols in aquatic sediment (Leeming et al. 2015). In this study, the coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) ratios exhibited obviously different values in these sediment with higher values in CH1 and lower in IIL1, but AC2 and BI had similar values (Table 2 and Fig. 2). Specifically, the ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) varied from 0.1 to 1.4 and 0.4 to 0.8 in the AC2 sediment core, with an average of 0.5 and 0.6, respectively. These two ratios ranged from 0.5 to 1.0 and 0.4 to 0.7 in the BI sediment core, with averages of 0.8 and 0.6, respectively. In the CH1 sediments, coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) displayed large variations from 3.1 to 10.8 and 0.9 to 1.0, with averages of 6.6 and 0.9, respectively. The IIL1 sediment core was characterized by small ranges with an average of 0.1 and 0.3 for these two ratios, respectively.

Fig. 2
figure 2

The ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in the AC2, BI, CH1 and IIL1 sediment collected in Ross Sea, Antarctica. The error bars are the 1σ standard deviations of all measured samples in the sediment profiles. The red area represents the ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in seal excrement in Antarctica

Full size image

The diversity of bacteria

The taxonomic compositions of bacteria in the AC2, CH1 and IIL1 sediment profiles are given in the supplementary materials (Figs. S4–S6). The dominant bacteria are different in these three explored sediment core. Specifically, Actinobacteriota (relative abundance varying from 6.1 to 15.6%), Bacteroidota (13.5 to 23.1%), Chloroflexi (6.1 to 19.5%), Firmicutes (12.5 to 38.8%), Proteobacteria (12.7 to 33.8%) dominated (relative abundance > 5%) in the AC2 sediment core (Supplementary Figure S4). In the CH1 sediment core, Bacteroidota (6.7 to 19.4%), Firmicutes (37.3 to 81.7%), Proteobacteria (3.5 to 28.3%) were the most abundant bacteria (Figure S5). Actinobacteriota (4.9 to 29.3%), Bacteroidota (2.9 to 35.2%) and Proteobacteria (45.2 to 86.9%) were predominant in the IIL1 sediment core (Fig. S6). We subsequently selected the sequences of ASVs identified to be the "Eubacterium_coprostanoligenes_group" in all the sequenced samples to build the phylogenetic tree (Fig. S7), and it indicates that the selected ASVs clustered together with the reference sequences of Eubacterium coprostanoligenes. Among these three explored sediment profiles, Eubacterium coprostanoligenes can be detected in the AC2 and CH1 sediment profiles, with relative abundance to total sequences ranging from 0 to 1.1%, and 0 to 1.2%, respectively. However, Eubacterium coprostanoligenes cannot be found in the IIL1 sediment profile.

Discussion

Distributions and sources of fecal sterols in sediments

Previous studies showed that BI and IIL1 sediment cores are significantly influenced by penguin guano according to inorganic elements associated with penguin guano (e.g., phosphorus), C/N ratios, and carbon isotope ratios of total organic matter (Liu et al. 2013; Wei et al. 2016; Jin et al. 2021; Zheng et al. 2022). For example, higher concentration of P, and lower δ13CTOC and C/N values indicate the input of animal guano-derived organic matter accounted for a larger proportion in the BI and IIL1 sediment profiles (Liu et al. 2013; Jin et al. 2021). Cholesterol is the most abundant sterol, comprising over 98% of the total sterols in fresh penguin guano (Leeming et al. 2015), consistent with our analysis of fresh Adélie penguin guano where cholesterol was higher than 99% (Table 1). According to comparative study, the concentrations of cholesterol in these four sediment profiles (AC2, BI, CH1, and IIL1) were similar with the fresh animal excrement and ornithogenic soil samples, while much higher than that of microbial mats, terrestrial moss, and soils (Supplementary Tables S2–S6). The above analyses are consistent with our field investigations that active colonies were observed at these four study area during our field investigations. Cholestanol in penguin guano is usually minor because penguins have no metabolic mechanism that can convert cholesterol to cholestanol (Martin et al. 1973). However, high abundance of cholestanol was also found in these four explored sediment profiles and likely originated from the microbial-mediated conversion of cholesterol in ornithogenic soil under aerobic conditions (Bull et al. 2002; Wang et al. 2007; Wisnieski et al. 2014). We assumed that, therefore, cholesterol in these four sediment profiles was mainly derived from animal excrement.

Coprostanol and epi-coprostanol were also found in the AC2, BI, CH1 and IIL1 sediment profiles (Table 2, Supplementary Figs. S2 and S3). According to many previous studies (Martins et al. 2002, 2014; Leeming et al. 2015), all three potential sources including human wastes, seal activities, and microbial hydrogenation of cholesterol to coprostanol by bacteria in anoxic conditions should be considered in Antarctic lacustrine sediment. The ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) can be used to distinguish the sources of fecal sterols in sediment (Fig. 2; Fattore et al. 1996; Leeming et al. 1996; Martins et al. 2014; Araújo et al. 2021). For example, the coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) values with the range of 0 to 1 and 0 to 0.5, respectively are inferred to be biogenic sources, while the ratio values more than 1 and 0.5 are strongly related to sewage input. The values of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) vary from 0 to 1 and 0.5 to 1, respectively, may indicate a microbial hydrogenation contribution. In this study, the ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) displayed large ranges in the four sediment profiles (Figs. 2, 3), indicating coprostanol may have different sources. We discuss the potential sources separately based on the ratios of fecal sterols.

Fig. 3
figure 3

The average values of coprostanol abundance (%), coprostanol/cholesterol ratio, and OI in four sediment profiles. The limit values of coprostanol/cholesterol on the Y-axis are different in the AC2, BI, CH1, and IIL1 sediment profiles. The error bars are the 1σ standard deviations of all measured samples in the sediment profiles

Full size image

Sources of coprostanol in the AC2 and BI sediment profiles

High abundances of coprostanol and epi-coprostanol in the AC2 and BI sediment profiles indicate that they may be strongly influenced by human sewage because these compounds are generally associated with human fecal matter (Bull et al. 2002). However, Beaufort Island and Adélie Cove are protected reserves and no human occupations have been occasionally established there except temporary camps, and these locations are also distant from research stations. Furthermore, the ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in AC2 and BI sediment cores were obviously lower than the values of sewage contamination (Fig. 2), indicating that coprostanol cannot be attributed to anthropogenic sources (Martins et al. 2014). Importantly, the bottom age of the BI sediment profile is ~ 1640 AD, suggesting no human activities at that time (Chen et al. 2015). Given these constraints, therefore, all results indicate that the high abundances of coprostanol and epi-coprostanol in the AC2 and BI sediment cores could not be attributed to anthropogenic sources.

The other likely source of the coprostanol would be derived from either marine mammals (seals) hauling out on the beach for breeding, resting, or molting, or microbial hydrogenation of sterols by bacteria in anoxic sediment. According to our previous studies, the extremely low δ13C values of organic matter in the BI sediment profile suggest that the total organic matter in the sediments mainly originated from animal excrement transported by melt-water erosion and runoff (Liu et al. 2013). However, seal hairs were not found throughout the AC2 and BI sediment profiles, though small numbers of Weddell and Elephant seals have occasionally been observed resting or molting on the beach (Hall et al. 2006; Mezgec et al. 2017; Koch et al. 2019). The AC2 sediment profile was collected in a small pond located on a highland that seals cannot access. Thus, we contemplated that the high levels of coprostanol and epi-coprostanol in the AC2 and BI sediment profiles were not caused by a large input of seal excrement, also supported by the distribution of sterols in fresh seal excrement. Leeming et al. (2015) reported that Elephant seal excrement had high cholesterol levels that accounted for more than 99% of the total sterols, with coprostanol and epi-coprostanol below 1%. This finding is similar to our results that the relative abundance of cholesterol is higher than 96% in seal excrement (Table 1). However, the average abundance of coprostanol is ~ 30% in the AC2 and BI sediment, and obviously higher than that of seal excrement. The different ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) between sediment and seal excrement infer that such high abundance of coprostanol may not be directly derived from seal excrement (Fig. 2). Additionally, some active Adélie penguin colonies were found at AC2 and BI sampling sites during our field investigations. On the contrary, the microbial-mediated bio-hydrogenation of cholesterol to coprostanol in deposited penguin guano should be considered.

At present, two major pathways for the conversion of cholesterol to coprostanol by gut microbiota include direct stereospecific reduction and an indirect pathway involving the intermediate formation of 4-cholelesten-3-one and coprostanone (Kriaa et al. 2019; Juste and Gérard 2021). For example, Juste and Gérard (2021) systematically summarized the associated active strains isolated from human and animal feces, and a hog sewage lagoon. Researchers have isolated a few bacteria belonging to the genus Eubacterium, Bifidobacterium, Clostridium, and Lactobacillus, which can convert cholesterol to coprostanol (Ren et al. 1996; Kriaa et al. 2019; Mukherjee et al. 2020). In this study, the Eubacterium coprostanoligenes can also be found throughout the AC2 sediment core (Supplementary Fig. S7). Additionally, the relative abundance of the Eubacterium coprostanoligenes has similar change pattern with the concentration of coprostanol in the AC2 sediment profile (Fig. 4). These results support our inference that cholesterol can be converted to coprostanol by specific bacteria in sediment.

Fig. 4
figure 4

The changes of cholesterol and coprostanol concentrations, coprostanol/cholesterol, Fe/Mn, OI values, and relative abundance of Eubacterium coprostanoligenes in the AC2 sediment profile

Full size image

Sources of coprostanol in the CH1 sediment profile

For the CH1 sediment profile, the most abundant fecal sterol is coprostanol with an average abundance of ~ 80% (Table 2), obviously higher than in animal excrement and the other three explored sediment profiles (AC2, BI and IIL1). The ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in the CH1 sediment profile are in the ranges of human fecal contamination (Fig. 2; Martins et al. 2014), indicating human sewage may be an important source of coprostanol. Wastewater from intensive human activities at the former research station at this location may have been discharged into ponds or catchments, thus accounting for the large amount of fecal sterols observed in the CH1 sediments. Many previous studies reported that abundant coprostanol was detected in Antarctic marine sediments near research stations, and relatively high ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) indicated it was primarily derived from sewage discharge (Martins et al. 2002, 2014; Leeming et al. 2015). The redox proxies (e.g., OI and Fe/Mn ratio) all indicate more suboxic sedimentary conditions in the CH1 sediment profile. Moreover, similar with AC2 sediment, sequences assigned to Eubacterium coprostanoligenes were also detected in the CH1 sediment core (Supplementary Figs. S7 and S8), suggesting microbial hydrogenation of cholesterol to coprostanol may also occur. Unfortunately, it is very difficult to distinguish the sources from sewage discharge and the microbial hydrogenation of cholesterol to coprostanol.

Sources of coprostanol in the IIL1 sediment profile

In the IIL1 sediment, the average abundance of coprostanol is lower than 10%, while cholesterol is the most abundant with an average abundance ~ 80% (Table 2). The IIL1 sediment was collected from a small pond, near many abandoned colonies of Adélie penguins on Inexpressible Island (Wei et al. 2016). Most importantly, many penguin remains including eggshells, bones, and feathers were found throughout the IIL1 sediment profile, indicating a strong influence from occupation by this species (Jin et al. 2021). Thus, it is reasonable to predict that high abundance of cholesterol is mainly derived from penguin guano. The ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in the IIL1 sediment profile fall between the ranges of seal excrement and penguin guano (Fig. 2), suggesting the small portion of coprostanol may be contributed by seal activities in this area. As reported by previous studies, the data of ancient DNA and 14C dating of elephant seal remains indicate this seal was present on Inexpressible Island from ~ 3270 to 380 BP, but likely only as occasional molting animals (Hall et al. 2006; Koch et al. 2019; Jin et al. 2021). We conclude that relative low abundance of coprostanol in the IIL1 sediment profile may have been contributed by seal use of the beaches for molting. These results are supportive of our sequencing data that the associated bacteria (e.g., Eubacterium coprostanoligenes) were not found in the IIL1 sediment profile. Furthermore, coprostanol is minor in the IIL1 sediment core, indicating that microbial hydrogenation of cholesterol to coprostanol did not occur to a noticeable extent.

The transformation of cholesterol to coprostanol in Antarctic lacustrine sediments

A number of laboratory experiments have demonstrated that cholesterol can be converted to coprostanol by gut microbiota in higher mammals (Kenny et al. 2020; Juste and Gérard 2021). In natural environments, the conversion of cholesterol to coprostanol was also reported in aquatic sediment under anoxic conditions (Gaskell and Eglinton 1975; Nishimura 1982; Jeng and Han 1996). For example, Nishumura (1982) reported that cholesterol could be transformed to coprostanol in anoxic aquatic sediment, which was controlled by the relative contribution of autochthonous organic matter. In Antarctic lakes influenced by animal excrements, higher organic matter inputs and degradations can consume oxygen, and form more suboxic sedimentary conditions (Lou et al. 2016). More suboxic conditions are necessary for this conversion to occur in lacustrine sediment (Nishimura 1982; Bull et al. 2002; Leeming et al. 2015). The Oxygen Index (OI), active iron Fe(III)/Fe(II) and Fe/Mn ratios are sensitive to the variability of redox sedimentary conditions in aquatic sediment (Haese et al. 1997; Jacob et al. 2004; Williams and Scherer 2004). For example, lower OI value, and higher Fe/Mn and Fe(II)/Fe(III) ratios usually indicate more suboxic sedimentary conditions. In this study, all these proxies in the AC2, BI, and CH1 sediment profiles indicated that these sediment were deposited under more suboxic conditions compared with IIL1 sediment profile (Fig. 3; Lou et al. 2016), which provided a potential sedimentary environment favorable for microbial hydrogenation (Gaskell and Eglinton 1975; Nishimura 1982). The concentrations of coprostanol actually show similar patterns of change with redox proxies in the AC2 and BI sediment profiles (Figs. 4 and 5), indicating more suboxic sedimentary conditions are suitable for microbial hydrogenation. By contrast, low abundance of coprostanol and low ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) in the IIL1 sediment profile with more oxic sedimentary condition are not suitable for microbial hydrogenation. Additionally, Hu et al. (2013) analyzed fecal sterols in a sediment profile (MB6) with less suboxic condition near penguin colonies at Cape Bird, Ross Sea. Abundant cholesterol and cholestanol were found in the top 32 cm, but coprostanol and epi-coprostanol were absent (Hu et al. 2013; Lou et al. 2016). Given these estimates, we conclude that high abundance of coprostanol was finally formed through microbial hydrogenation of cholesterol in the AC2 and BI sediment profiles under more suboxic conditions.

Fig. 5
figure 5

The changes of cholesterol and coprostanol concentrations, δ13CTOC, coprostanol/cholesterol, Fe(III)/Fe(II) and OI values in the BI sediment profile. The data of δ13CTOC are cited from Liu et al. (2013). The data of OI and Fe(III)/Fe(II) was cited from Lou et al. (2016). The concentration of cholesterol was cited from Zheng et al. (2022)

Full size image

The concentrations of coprostanol exhibit large ranges versus depth in the AC2 and BI sediment profiles (Figs. 4 and 5), suggesting different microbial hydrogenation strength of cholesterol to coprostanol. We found that coprostanol concentrations showed similar change patterns with guano input strength and redox sedimentary conditions (Figs. 4 and 5). High input intensity of guano can carry more organic matter and nutrition into ponds, resulting in an anoxic sedimentary condition in the AC2 and BI sediment profiles (Lou et al. 2016). These favorable conditions can cause the occurrence of bio-conversion of cholesterol to coprostanol in lacustrine sediment. Our results suggest that the higher concentrations of coprostanol are in line with more suboxic conditions and stronger guano input in the AC2 and BI sediment profiles, indicting more favorable sedimentary conditions for this conversion. However, isotope labeling cultures should be performed to further verify the transformation of cholesterol to coprostanol in natural sediment.

Conclusion

Abundant C27 fecal sterols (cholesterol, cholestanol, coprostanol, and epi-coprostanol) were found in four lacustrine sediment profiles (AC2, BI, CH1, and IIL1) that were strongly influenced by animal excrement. Cholesterol is the most abundant sterol in the AC2, BI, and IIL1 sediment profiles, while coprostanol is only predominant in CH1. Based on the ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol), the cholesterol in the IIL1 sediment profile is mainly derived from penguin guano, while a small part of coprostanol is contributed by seal feces at Inexpressible Island. However, abundant coprostanol is found in the AC2 and BI sediment profiles, which were strongly influenced by penguin guano. The ratios of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) and lack of seal hairs strongly suggest that coprostanol in the AC2 and BI sediment profiles was mainly derived from the microbial hydrogenation of cholesterol from penguin guano. However, in the CH1 sediment, the high values of coprostanol/cholesterol and coprostanol/(coprostanol + cholestanol) indicate that coprostanol is likely related to human sewage due to research activities from 1968 to 2006 at Hallett Station.

Compared with the isolated cholesterol-reducing bacterium, we found the cholesterol-reducing bacterium (Eubacterium coprostanoligenes) present throughout the two more suboxic sediment profiles (AC2 and CH1), but absent in the sediment profile (IIL1) with minor amounts of coprostanol. Our data indicated that cholesterol was likely converted to coprostanol and epi-coprostanol via microbial hydrogenation processes by bacteria under anoxic aquatic systems in Antarctica. Therefore, caution should be taken when using coprostanol to infer seal activities and human waste in Antarctic anoxic ponds or lakes.