In this study, we reanalysed a global (n = 916) and south of 35° S (n = 289) core-top isoGDGT datasets, by adding data from Jaeschke et al. (2017) and Lamping et al. (2021) to the calibration dataset presented in Tierney and Tingley (2014) (Fig. 1). SST data for each core-top site were derived from the World Ocean Atlas 2009 (WOA09) 1° × 1° gridded product (Locarnini et al., 2010). Additional environmental parameters that potentially influence isoGDGT composition – including annual mean and seasonal seawater temperatures (0–1400 m depths), salinity (0–1400 m depths), and oxygen saturation (0–1400 m depths) – were derived from the World Ocean Atlas 2018 (WOA18) 0.25° × 0.25° gridded product (Boyer et al., 2018).

Previously reported and newly generated sedimentary isoGDGT records from multiple sites around the ACC were used to evaluate the applicability of isoGDGTs as a palaeoceanographic proxy in the Southern Ocean (Fig. 1). New isoGDGT records were obtained from sediment cores U1538 (57.43° S, 43.35° W, 3131 m water depth) and U1537 (59.11° S, 40.91° W, 3713 m water depth) collected in the northern and southern Scotia Sea near the SBF during the International Ocean Drilling Program Expedition 382 (Weber et al., 2019). The sites are located in the Dove Basin in the southern Scotia Sea, where sedimentation is influenced by the transport of eastward-flowing ACC and northward-flowing Weddell Sea Deep Water (Weber et al., 2019). Additionally, we analysed previously reported isoGDGT records in sediment cores MD11-3357 (44.68° S, 80.43° E, 3349 m water depth), MD12-3394 (48.38° S, 64.58° E, 2320 m water depth) and MD11-3353 (50.57° S, 68.39° E, 1568 m water depth) collected from the southeastern Indian Ocean (Ai et al., 2020; 2024).

Organic compounds were extracted from approximately 2 g of freeze-dried sediment using a DIONEX Accelerated Solvent Extractor 200 with dichloromethane/methanol (9:1) at a temperature of 100 °C and a pressure of 1000 psi. The extracts were initially separated into neutral and acidic fractions by aminopropyl silica gel column chromatography with dichloromethane/2-propanol mixture (2:1). The neutral fraction was further separated into two fractions (N1-3 and N4) by silica gel column chromatography. The N4 fraction, which contains GDGTs, was dissolved in hexane/2-propanol (99:1) and filtered through 0.45 µm filters.

GDGTs in U1537 and U1538 samples were identified and quantified using high-performance liquid chromatography-mass spectrometry (HPLC-MS) with an Agilent 1260 HPLC system coupled to 6130 quadrupole mass spectrometers. Separation was achieved with a Prevail Cyano column (2.1 × 150 mm, 3 µm; Grace Discovery Science, USA) maintained at 30 °C following the method of Hopmans et al. (2000) and Schouten et al. (2007). The analytical conditions were as follows: flow rate 0.2 mLmin-1, isocratic with 99 % hexane and 1 % 2-propanol for the first 5 min followed by a linear gradient to 1.8 % 2-propanol over 45 min. Detection was achieved with an atmospheric pressure chemical ionization-MS (APCI-MS). The spectrometer was run in two different selected ion monitoring modes (m/z 743.8, 1018, 1020, 1022, 1032, 1034, 1036, 1046, 1048, 1050, 1292.3, 1296.3, 1298.3, 1300.3, and 1302.3). Following the literature of Hopmans et al. (2004), compounds were identified by comparison of mass spectra and retention times. GDGTs were detected and quantified by integrating the peak area in the (M+H)+ chromatogram with a comparison to the peak area of an internal standard (C46 GTGT) in the (M+H)+ chromatogram, according to the method of Huguet et al. (2006). By comparing the peak areas of isolated Cren, GDGT-0, and C46 GTGT in known amounts, the correction value of ionization efficiency between GDGTs and the internal standard was determined (Schouten et al., 2007). To monitor the changes in the ionization efficiency, a mixture of C46 GTGT as the working standard and the GDGTs extracted and purified from an East China Sea sediment was inserted in the routine analysis every 20 samples. The standard deviations of Cren, GDGT-1, GDGT-2, and GDGT-3 in replicate LC/MS analysis were 1 %, 1 %, 3 %, and 2 % in sediment samples, respectively.

A number of indices and calibrations have been proposed to reconstruct ocean temperature and assess non-thermal effects based on isoGDGT composition. The isoGDGT-based indices and methods used in this study are described below. The TEX86 index was first proposed by Schouten et al. (2002) and is defined as the following Eq. (1), where the bracketed GDGTs represent the relative abundance of each compound. 1TEX86=[GDGT-2]+[GDGT-3]+[Cren′][GDGT-1]+[GDGT-2]+[GDGT-3]+[Cren′]

The conversion of the TEX86 index to temperature was initially achieved using a linear relationship (Schouten et al., 2002). A range of calibrations have subsequently been developed (Kim et al., 2010; Liu et al., 2009; Schouten et al., 2002). This includes a spatially varying calibration model for TEX86 based on a Bayesian approach, BAYSPAR (Tierney and Tingley, 2014). The TEX86L index, was also proposed as more suitable for reconstructing temperatures below 15 °C and is defined as the following Eq. (2) (Kim et al., 2010). 2TEX86L=log⁡10[GDGT-2][GDGT-1]+[GDGT-2]+[GDGT-3]

The OPTiMAL calibration, a machine learning approach, uses all six isoGDGTs in global core-top data as training data to estimate SST (Dunkley Jones et al., 2020). The training dataset (n = 914) used in this study consists of the global “Op1” dataset (Dunkley Jones et al., 2020) and additional core-top data from the Southern Ocean (Jaeschke et al., 2017; Lamping et al., 2021).

IsoGDGTs are biosynthesized not only by Nitrososphaera but can also be derived from other sources including methanogenic, methanotrophic and terrigenous archaea. A number of isoGDGT-based indices have been proposed to assess the non-thermal effects caused by exogenous isoGDGT inputs (Blaga et al., 2009; Sinninghe Damsté et al., 2012; Taylor et al., 2013; Pearson and Ingalls, 2013). The methane index (MI) has been proposed as an indicator of post-depositional methanotrophic archaeal GDGT input if the value exceeds 0.3 and is defined as follows (Zhang et al., 2011): 3MI=[GDGT-1]+[GDGT-2]+[GDGT-3][GDGT-1]+[GDGT-2]+[GDGT-3]+[Cren]+[Cren′]

The relative abundance of Cren′ to Cren, expressed as fcren (Eq. 4) is interpreted as reflecting non-temperature-related influences, particularly community changes in Nitrososphaera that produce differing amounts of Cren′ (O'Brien et al., 2014; Pitcher et al., 2010). This is relevant when fcren exceeds 0.25 (O'Brien et al., 2014): 4fcren=[Cren′][Cren′]+[Cren]

Additionally, the ratio of GDGT-2 and GDGT-3 ([2]/[3]) has been suggested to reflect a contribution from archaea living deeper in the water column, especially when the values exceed 5 (Taylor et al., 2013). 5[2]/[3]=[GDGT-2][GDGT-3]

%GDGT-0 has been proposed as an indicator of the contribution of methanogenic archaea when the value exceeds 67 %, defined as follows (Sinninghe Damsté et al., 2012): 6[%GDGTtext-0]=[GDGT-0][GDGT-0]+[Cren]×100

As shown in Fig. S2, correlations between the conventional isoGDGT indices (TEX86 and TEX86L) and WOA-derived in situ SST are weak in the low temperature range (< 5 °C), calling into question the applicability of the conventional indices as SST proxies in the polar region. Therefore, we conducted principal component analysis (PCA) on the fractional abundances of six isoGDGTs from sites located south of 35° S (n = 289; Fig. 2a) to evaluate the factors specific to the production of isoGDGTs in the Southern Ocean. For comparison, PCA was also performed on a global core-top dataset (n = 916; Fig. 2b). In both the Southern Ocean and global datasets, the first two principal components explain over 90 % of the total variance, with PC1 accounting for approximately 70 % (Table 1).

In the Southern Ocean dataset, PC1 exhibits strong positive loadings for GDGT-1 (0.69) and GDGT-2 (0.56), suggesting that these two compounds are the primary contributors to this component (Table 1, Fig. 2a). In contrast, GDGT-0 displays a negative loading (-0.23) on PC1, which is also observed in the global PCA. However, the primary contributors to PC1 in the global dataset differ, with GDGT-2 and Cren′ showing the highest loadings in contrast to the Southern Ocean dataset (Table 1, Fig. 2b).

For PC2, the Southern Ocean dataset is primarily influenced by GDGT-0 and Cren, which exhibit opposing loadings: GDGT-0 loads negatively, while Cren loads positively. In contrast, global PC2 is mainly driven by GDGT-1 and Cren, which also load in opposite directions. These differences highlight that the distribution of isoGDGTs in the Southern Ocean is not entirely consistent with that observed in the global dataset.

To determine what PC1 represents in both global and Southern Ocean datasets, we compared it with the environmental parameters that potentially influence isoGDGT composition, including SST, salinity, and oxygen saturation at surface (0 m) depth (Qin et al., 2015; Schouten et al., 2013). Five sites in the North Pacific were excluded from the global core-top dataset, as they formed distinct clusters in the PCA and exceeded the thresholds of the screening methods (MI and %GDGT-0) (Fig. 2b). In the global dataset, PC1 exhibits the strongest correlation with SST (Table 2: R = 0.89), supporting the use of isoGDGT distributions as a palaeotemperature proxy. In the Southern Ocean dataset, PC1 also shows the highest correlation with SST among the environmental variables; however, the correlation is considerably weaker than in the global dataset (Table 2; R = 0.69). This suggests that, in contrast to the global dataset, isoGDGT distributions in the Southern Ocean may be substantially influenced not only by temperature but also by other environmental factors.

We further examined relationship of PC1 with non-thermal GDGT indices such as MI (Zhang et al., 2011), fcren (O'Brien et al., 2014), [2]/[3] (Taylor et al., 2013), and %GDGT-0 (Sinninghe Damsté et al., 2012). Both the Southern Ocean and global PC1 show strong positive correlations with MI and fcren (Table 2), with values remaining below the critical thresholds of 0.3 and 0.25, respectively. This indicates that anomalous isoGDGT distributions are not apparent, suggesting the contribution of post-depositional methanotrophic archaea is negligible.

One of the notable distinctions between the global and Southern Ocean datasets lies in the relation between PC1 and [2]/[3]. The [2]/[3] ratio is used to differentiate between contributions from “shallow” (0–200 m depth) and “deep” (> ∼ 1000 m depth) clades of archaea, with higher values reflecting increased contributions from archaeal communities residing at deeper water depths (Rattanasriampaipong et al., 2022). Specifically, [2]/[3] values above 5 indicate a significant contribution from deeper-dwelling archaea (Taylor et al., 2013). While global PC1 shows only a weak correlation with [2]/[3] (Table 2; R = 0.38), Southern Ocean PC1 exhibits a much stronger correlation (Table 2; R = 0.74), with a steeper slope, including several values exceeding this threshold (Figs. S3a and S4a in the Supplement). This association implies that isoGDGTs in sediment samples in the Southern Ocean are more strongly influenced by archaea living in the deeper ocean. The spatial distribution of [2]/[3] exhibits a latitudinal pattern: lower values occur south of the ACC, whereas northern sites frequently have higher values exceeding the threshold (> 5). If isoGDGT distributions were controlled solely by temperature, archaea at lower-latitude sites would produce isoGDGTs with more rings, leading to lower [2]/[3] values. However, the pattern we observe here appears counterintuitive, with northern sites clustering at higher values.

Several mechanisms may explain the observed spatial variability in [2]/[3] values. First, basin bathymetry provides control on [2]/[3] distributions. Our finding that [2]/[3] ratios correlate with site water depth (R = 0.74) suggests that in deeper basins, the relative contribution from deep-dwelling archaea is increased. This bathymetric effect is particularly pronounced at northern Southern Ocean sites, where core-top samples predominantly come from deep basin locations (mean depth ∼ 3400 m). Conversely, sites on the shallower Antarctic continental shelf and slope maintain lower [2]/[3] ratios, likely due to shallow archaeal clades dominating the water column.

Additionally, the relationship between export depth and the depth of maximum isoGDGT production differs within the water column. Previous study has shown that peak isoGDGT production occurs at the base of the photic zone and/or below the nitracline (Hernández-Sánchez et al., 2014; Spencer-Jones et al., 2021; Wuchter et al., 2005). Nitracline depths in the Southern Ocean vary with latitude, with northern sites, particularly in the South Pacific subtropical gyre region, exhibiting deeper nitraclines (several hundred meters to 1000 m) due to strong permanent stratification and Ekman downwelling at gyre centres (Dai et al., 2023; Feucher et al., 2019). These deeper nitraclines could increase contributions from deeper archaeal clades to seafloor sediments. In addition, the oligotrophic conditions at northern subtropical sites result in low surface particle flux (Dai et al., 2023). Since export of isoGDGTs depends on sinking particles (e.g., marine snow, fecal pellets, phytoplankton aggregates) that incorporate lipids (Huguet et al., 2006; Wuchter et al., 2005), weak surface productivity reduces the contribution of surface-dwelling archaeal signals, allowing GDGTs from deeper archaeal communities to contribute proportionally more to the sedimentary record, thereby elevating [2]/[3] values. In contrast, southern sites are characterized by upwelling of CDW, which brings nutrients to the upper water column and creates shallower nitraclines with enhanced surface productivity (Carter et al., 2008), leading to lower [2]/[3] values. These mechanisms can potentially explain the stronger correlation between PC1 and [2]/[3] observed in the Southern Ocean compared to the global dataset, introducing a significant non-thermal influence on sedimentary isoGDGT distributions across the Southern Ocean.

The relationship between PC1 and %GDGT-0 is also markedly different between the global and Southern Ocean datasets. In the global dataset, %GDGT-0 exhibits a strong negative correlation with PC1 (Table 2; R = -0.87), whereas in the Southern Ocean, this correlation is much weaker (Table 2; R = -0.25). Interestingly, the slope of the relationship changes across the PF: sites south of the PF exhibit a positive slope, while those to the north follow the negative trend observed globally (Fig. 3a; see also Fig. S4a). A similar divide is evident in the relationship between PC1 and SST in the Southern Ocean (Fig. 3b), where the data form two distinct clusters corresponding to locations north and south of the PF. In Cluster 1, which primarily represents sites in the Antarctic Zone (south of the PF), the correlation between PC1 and SST is stronger (R = 0.77) than that observed for the full Southern Ocean dataset (R = 0.69). These findings indicate that the distribution of isoGDGTs in the Southern Ocean differs between the regions north and south of the PF, with SST exerting a stronger influence on isoGDGT distributions in the Antarctic Zone.

In the Southern Ocean, water column structure and associated processes such as vertical mixing and upwelling of CDW vary in the different zones delimited by the oceanic fronts, with a particularly pronounced transition at the PF (Carter et al., 2022). To explore how these oceanographic features relate to isoGDGT distributions, PC1 values from the Southern Ocean were compared across different water masses using two classification schemes: one dividing the ACC into south of the ACC (south of the SBF), centre of the ACC (between the SAF and SBF), and north of the ACC (north of the SAF) (Fig. 4c), and another based on position relative to the PF (north and south of the PF) (Fig. 3c). A boxplot of PC1 values separated by the three frontal zones (Fig. 4a) highlight distinct distributions for each zone, suggesting significant differences associated with zonal water mass structure. However, the most pronounced separation occurs between sites south and north of the PF. Furthermore, the scatter plot of PC1 vs. isoGDGT-based indices shows significant linear correlations, with data points transitioning from south of the ACC (lower left) to north of the ACC (upper right), with centre of the ACC forming an intermediate cluster (Fig. S3a). These results suggest that the spatial variability in isoGDGTs in the Southern Ocean is primarily caused by zonal water mass transitions across the ACC, with the PF representing a particularly sharp boundary. This is consistent with a previous study which shows a shift in composition of intact polar lipid GDGTs in the Scotia Sea across well-defined fronts (Spencer-Jones et al., 2021).

The PCA of isoGDGTs in the Southern Ocean core-top samples showed that PC1 (70.3 % of the variance) primarily reflects temperature and export depth associated with the zonal water mass properties along the north-south transect across the PF. Thus, the isoGDGTs which strongly correlate with PC1 could serve as a valid indicator of the zonal water masses in the ACC region. Based on the PCA analysis, we propose a novel isoGDGT-based index as a water mass tracer, composed of GDGT-0, GDGT-1, and GDGT-2, which exhibit significant correlation with PC1. We term this the Antarctic IsoGDGT Zonal (AIZ) index, defined as: 7AIZ=GDGT-1+[GDGT-2][GDGT-0]+[GDGT-1]+[GDGT-2]

This new index has a strong linear relationship with PC1 (R2 = 0.98, n=289), with higher values from samples north of the PF (Fig. 4b). The index values display a clear gradient across the PF, with a threshold of 0.14 (± 0.03 standard deviation, SD) determined from the interquartile range of sites at the present-day PF position (Fig. 5).

We propose that the AIZ index responds to changes in archaeal community composition delineated by water mass changes across the PF. Genomic studies in the South Pacific have identified that oceanographic features, such as wind-driven circulation at the surface, are the primary drivers of prokaryotic richness and community diversity patterns (Kolody et al., 2025; Raes et al., 2018). North of the PF, archaeal richness increases northward and peaks at the STF border, which also acts as an ecological boundary (Raes et al., 2018). South of the PF, the diversity of the archaeal community is low, possibly due to the extremely cold and harsh conditions that may have acted as a habitat bottleneck (Alonso-Sáez et al., 2011). Genomic analysis in waters south of the PF shows “Candidatus Nitrosopumilus maritimus” dominates the Nitrososphaera phylum (Hernández et al., 2015; Kim et al., 2014; Sow et al., 2022). Moreover, Spencer-Jones et al. (2021) conducted a PCA of intact polar lipid GDGT compositions in the Amundsen and Scotia Seas, along with previously published clusters of Nitrososphaera (Bale et al., 2019) and found clustering within the Nitrosopumilales group in both regions due to the high relative abundances of GDGT-0. We suspect that the AIZ index captures these changes in archaeal community composition, specifically reflecting the dominance of cold-adapted Nitrosopumilales south of the PF in contrast to more diverse communities northward. Because these community shifts are tied to water mass boundaries, the AIZ index can serve as a good indicator for reconstructing past PF movements.

The difference between the traditional isoGDGT-based and AIZ-based approaches is likely attributed to the exclusion or inclusion of GDGT-0. GDGT-0 has traditionally been excluded from temperature indices. The fractional abundances of GDGT-0 are generally much higher than those of other GDGTs, which would otherwise overpower the index calculations (Schouten et al., 2002; Kim et al., 2010). Thus, one of the reasons for this exclusion is based on mathematical rather than ecological considerations. However, previous studies suggest that GDGT-0 does hold temperature information (Dunkley Jones et al., 2020; Kim et al., 2010; Zhao et al., 2025). The fractional abundance of GDGT-0 shows a strong correlation with SST in the global core-top dataset (R2 = 0.72,n = 894), indicating that GDGT-0 is intrinsically associated with growth temperature across a wide temperature range. In fact, the production of GDGT-0 increases with decreasing water temperature (Kim et al., 2010). This is explained by the physiological necessity to reduce the number of cyclopentane rings in the archaeal membrane lipids to maintain membrane fluidity at low temperatures (Fietz et al., 2020; Gabriel and Chong, 2000; Schouten et al., 2002). Moreover, Dunkley Jones et al. (2020) found GDGT-0, along with GDGT-3 to be the most informative GDGTs for predicting temperature using Gaussian process regression. Therefore, the inclusion of GDGT-0 becomes particularly important at high latitudes, where the relative abundance of GDGT-3 is substantially reduced. Another reason that GDGT-0 has traditionally been excluded from TEX86 is the potential influence of factors other than temperature on its abundance in sediments. There are alternative sources of GDGT-0, such as methanogenic and methanotrophic archaea living in anoxic sedimentary environments (Pancost et al., 2001; Zhang et al., 2011). However, methanotrophic archaea produce not only GDGT-0 but also GDGTs-1, -2, and -3 (Pancost et al., 2001; Schouten et al., 2013). We acknowledge the potential impact of non-Nitrososphaera sources when using GDGT-0. Nevertheless, by combining screening tests (%GDGT-0 and MI), these non-thermal influences can be minimized. We note that the Southern Ocean dataset we used here to establish the AIZ index was all screened by using %GDGT-0 and MI.

The weak correlation between temperature and conventional isoGDGT-based indices (TEX86 and TEX86L) in the polar region questions their applicability as reliable temperature proxies in the region (Fietz et al., 2016; Kim et al., 2008). On the other hand, the AIZ index exhibits a relatively strong correlation (R = 0.77) with SSTs in the Antarctic Zone, specifically south of the PF (Fig. 3b) where the source of GDGTs is predominantly the Nitrosopumilales group. This finding suggests that AIZ has a potential as a reliable proxy for reconstructing ocean temperature in the Antarctic Zone.

It has been reported that the seasonality and depth of isoGDGT production is regionally dependent in the Southern Ocean (Church et al., 2003; Park et al., 2019; Sow et al., 2022; Spencer-Jones et al., 2021). To evaluate the potential of AIZ as a temperature proxy, we examined correlations between AIZ values and ocean temperatures at different depths (0–1200 m) across different seasons (summer, autumn, winter, spring, and annual mean) derived from the WOA18 0.25° × 0.25° gridded product (Boyer et al., 2018). The results show that the correlation varies considerably with depth but is less affected by seasonal changes. Significant correlations (R2 = 0.63–0.82) were observed at depths below 200 m in all seasons, peaking at 400 m depth (R2 = 0.77–0.82) (Table 3, Fig. 6). Notably, the correlation of the logarithmic calibration (R2 = 0.66–0.90) is higher than that of the linear calibration at depths below 200 m (Table 3, Fig. 6). This pattern of improved correlation with non-linear calibrations has been recognised previously for conventional isoGDGT indices in Southern Ocean core-top datasets (Park et al., 2019). Correlations between AIZ values and both salinity and oxygen concentration at subsurface depths across seasons were also evaluated, but these correlations were much weaker than that between AIZ and ocean temperature (R2 = 0.0001–0.45).

Our results, which show the highest correlation of the AIZ index with subsurface temperatures, suggest that isoGDGTs in the Southern Ocean are primarily produced at subsurface depths, potentially associated with the archaea inhabiting CDW. CDW is pervasive throughout the water column but centred at approximately 500 m depth south of the PF (Holland et al., 2020), consistent with the 400 m depth where the strongest correlation is observed. Subsurface production of isoGDGTs from Nitrososphaera in the Southern Ocean has also been suggested by previous work, including 16S rRNA gene (Church et al., 2003; Kalanetra et al., 2009; Sow et al., 2022), living archaea (Spencer-Jones et al., 2021), sediment trap studies (Park et al., 2019) and sedimentary records (Etourneau et al., 2019; Ho and Laepple, 2016; Kim et al., 2012; Lamping et al., 2021; Liu et al., 2020). Furthermore, it has been reported that there are significant reductions in Nitrososphaera abundance above 100 m depth in the Southern Ocean (Signori et al., 2014) possibly due to light inhibition (Merbt et al., 2012), which is consistent with the decline in correlation above 200 m depth. These lines of evidence indicate that the AIZ index most likely reflects mesopelagic (200–1000 m depth) water temperature. The calibration equations based on annual mean temperature at 400 m depth (subsurface temperature: subST) where the strongest correlations (both linear and logarithmic) were obtained, are as follows: 8Linear: subST=24.17×AIZ-1.45(R2=0.81,n=134,p<0.0001)9Logarithmic: subST=2.08ln⁡(AIZ)+6.11(R2=0.88,n=134,p<0.0001)

It is important to note that AIZ index-based palaeothermometry is only applicable within the Antarctic Zone. We suggest a threshold value of 0.14 (± 0.03 SD) based on the present PF position. Below this threshold, deposition occurs south of the PF, where isoGDGTs are predominantly produced by Nitrosopumilales group, and isoGDGT composition shows a strong correlation with temperature. In contrast, above 0.14, deposition occurs north of the PF, where different archaeal communities result in a distinct temperature-AIZ relationship, and the calibration cannot be reliably applied. If a sediment core reconstruction contains intervals where AIZ values fluctuate above and below 0.14, temperatures should only be reconstructed for intervals below this threshold.

To further test the validity of the AIZ index as a water temperature proxy in the Antarctic Zone, we compared it with the OH-isoGDGT derived indices (TEX86OH, RI-OH, RI-OH′, OHc, and %OH; Table S1 in the Supplement). While OH-isoGDGTs are not available for all datasets, we used dataset from Lamping et al. (2021), which contains both isoGDGTs and OH-isoGDGTs (n = 66), to compare different temperature proxies. We compared SST (0 and 0–200 m) derived from OH-isoGDGT-based indices with subST (400 m) derived from the AIZ index. The AIZ index shows the strongest correlations with TEX86OH (R2 = 0.94) and OHc (R2 = 0.94), while other OH-isoGDGT-based indices also demonstrate good correlations listed as follows: RI-OH (R2 = 0.79), RI-OH′ (R2 = 0.68), and %OH (R2 = 0.64) (Fig. S5a–e in the Supplement). These strong correlations between the AIZ index and OH-isoGDGT-based indices, particularly with the recently developed TEX86OH, suggest that both proxies capture similar temperature patterns. However, when comparing calibration performance, the AIZ-derived subST400m shows better agreement with WOA-derived and modelled annual mean subST at 410 m (provided by Lamping et al., 2021) than TEX86OH-derived SST0–200m (Fig. S5f–g), highlighting the importance of regionally-specific calibrations in polar regions. We agree with Bijl et al. (2025) and Varma et al. (2024) that OH-isoGDGTs are valuable for temperature reconstruction, particularly in regions with temperatures < 15 °C, and recommend measuring OH-isoGDGTs in future studies. Nevertheless, the AIZ index and its calibration offer a valuable tool for the many instances where isoGDGT data are already published or measured, but OH-isoGDGTs are not available.