Total organic residues obtained from palynological preparations of eight Waipawa and two Whangai samples from the Taylor White section were solvent extracted to remove any bitumen present and then processed by density fractionation with the aim of identifying the palynofacies source or sources for 13C enrichment in Waipawa organofacies (Table S3). The preparations (3–597 mg) were sieved into grain-size fractions < 6 and ≥ 6 µm. Both grain-size fractions were processed by density separation using sodium polytungstate (Na6[H2W12O40]; high-purity SPT-0, TC-Tungsten Compounds GmbH, Germany) in deionized water. Five to eight density fractions per grain-size fraction were obtained, from < 1.2 to > 1.5 g cm-3. Palynofacies analysis was carried out on 86 density fractions in the ≥ 6 µm suite. Carbon isotope analysis was carried out on 77 samples in this suite and 10 samples in the < 6 µm suite (Table S3).

To investigate the source and composition of OM and to help reconstruct depositional and environmental conditions, lipid biomarkers and the carbon isotope composition of the total saturated and aromatic hydrocarbon fractions in solvent extracts (bitumen) from the Taylor White section were previously analysed at Applied Petroleum Technology (APT) in Oslo, Norway, with the methods and data reported by Naeher et al. (2019). Only the carbon isotope values of the total saturated and aromatic fractions (Table S4) and representative chromatograms are presented in this paper.

Compound-specific carbon isotope analyses of selected isoprenoids (pristane and phytane), n-alkanes (nC18–nC33) (Table S4) and fatty acids (Table S6) were undertaken in the Organic Geochemistry Laboratory at GNS Science and the Organic Geochemistry Unit (OGU) at the University of Bristol, UK. Samples from the Taylor White section were prepared using the analytical procedures reported in Naeher et al. (2012, 2014) with some modifications. In brief, 7–40 g powdered rock was extracted (four times) with dichloromethane (DCM) / methanol (MeOH) (3 : 1, v/v) by ultrasonication for 20 min each time. Elemental sulfur was removed by activated copper. The total lipid extracts (TLEs) were divided into saturated and aromatic hydrocarbon and polar compound fractions via liquid chromatography over silica columns using n-hexane, n-hexane / DCM (7 : 3, v/v), and DCM / MeOH (1 : 1, v/v), respectively. An aliquot of the polar fraction was derivatized with BSTFA (N,O-bis(trimethylsilyl)trifluoroacetamide) (Sigma Aldrich) for 1 h at 80 ∘C prior to analysis.

For compound-specific carbon isotope analyses of pristane (Pr), phytane (Ph) and n-alkanes (Table S4) we undertook molecular sieving (Dawson et al., 2005; Grice et al., 2008; Aboglila et al., 2010) of free and desulfurized saturated fractions of samples from the Taylor White section. Saturated hydrocarbon fractions dissolved in cyclohexane were added to an activated 5 Å molecular sieve (Alltech) and heated for 8 h at 80 ∘C. Branched and cyclic compounds were recovered by extraction (five times) with cyclohexane. n-Alkanes were recovered by dissolution of the sieve with 30 % HF, followed by neutralization with a saturated NaHCO3 solution. The resulting fractions were analysed using an Isoprime 100 GC–C-IRMS system at the University of Bristol. Injection volume was 1 µL onto to a Zebron-I nonpolar column (50 m × 0.32 mm inner diameter, 0.10 µm film thickness). The GC oven programme was 3 min hold at 70 ∘C, heating to 130 ∘C at 20 ∘C min-1 and then to 300 ∘C at 4 ∘C min-1, and a final hold at 300 ∘C for 25 min. Samples were measured in duplicate and δ13C values converted to Vienna Pee Dee Belemnite (VPDB) by bracketing with CO2 of known δ13C value. Instrument stability was monitored by regular analysis of an in-house fatty-acid methyl ester standard mixture.

The fatty acids from the mid-Waipara samples were extracted and analysed using a different extraction and separation protocol. Powdered samples were placed in pre-extracted cellulose thimbles and extracted under reflux using a Soxhlet apparatus for 24 h with DCM / MeOH (2 : 1 v/v) as the organic solvent. The resulting TLEs were separated using an aminopropyl (NH2) solid phase extraction (SPE) column by elution with DCM/isopropanol (2 : 1 v/v; neutral fractions), followed by 2 % (by volume) acetic acid in diethyl ether (acid fractions). The columns were glass cartridges containing 500 mg of silica-bonded stationary phase, manufactured by Isolute^®. Acid fractions were then methylated using a BF3 / MeOH complex (14 % w/v; 100 µL; 60 ∘C for 30 min). After cooling to room temperature, ∼ 1 mL of double-distilled water was added and then extracted with ∼ 2 mL of DCM. The extracts were passed through a short glass pipette column packed with pre-extracted glass wool and sodium sulfate (Na2SO4) (to remove residual water). A further two repeat extractions were performed on each fraction (eluted into the same vial), resulting in a combined extract, which was then dried under N2. The methylated acid fractions were then silylated at 70 ∘C for 1 h.

For compound-specific carbon isotope analyses of the fatty acids (Table S6) GC–C–IRMS was conducted using a Hewlett Packard 6890 gas chromatograph connected to a Thermoquest Finnigan Delta plus XL spectrometer, via a GC III combustion interface (comprising Cu, Pt and Ni wires within a fused alumina reactor at a constant temperature of 940 ∘C). GC conditions were as described above. Duplicate analyses were conducted for each sample, with the mass balance of values corrected for the addition of a methyl group and reported in standard delta (‰) notation relative to VPDB. Analytical precision, based on replicate analysis of a standard of mixed fatty-acid methyl esters (FAMEs), is < ±0.5 ‰.

The Waipawa organofacies was defined by Hollis et al. (2014) as a distinctive, organic-rich marine facies in which the organic matter is enriched in 13C. The lithology is typically mudstone, with varying proportions of detrital sand, biogenic silica and glauconite. The additional samples and sites examined in this study confirm that, irrespective of background sediment type, Waipawa organofacies is readily identified by a combination of relatively elevated TOC values, typically > 1 wt % and up to about 15 wt %, and enriched δ13COM values of -24 ‰ to -17 ‰ (Fig. 3, Table S2). There is a positive correlation between TOC and δ13COM in all sections examined, even in sections such as Glendhu Rocks where maximum TOC is low in comparison with other sites (i.e. < 0.5 wt %).

Having established that the 13C enrichment appears to be linked primarily to the dominance of terrestrially derived degraded phytoclasts in the Waipawa OM (Fig. 5), we now consider how and to what extent OM degradation or indeed preservation processes within the broader depositional environment might account for the 13C enrichment. Only by accounting for potential processes of 13C enrichment during OM transportation, deposition and early diagenesis is it possible to identify any residual enrichment that may be related to a drawdown in atmospheric CO2 levels.

To circumvent the complications of lignin degradation and source mixing that affect the δ13C values of the bulk OM and the saturated and aromatic hydrocarbon fractions, we now focus on compound-specific carbon isotope analysis of sediments from the Taylor White (Fig. 9a, b) and mid-Waipara sections (Fig. 10a). For Taylor White, we analysed pristane, phytane and C18–C33 n-alkanes (δ13CPr, δ13CPh, δ13C18, etc.; Table S5). For mid-Waipara, we analysed the C16–C32 n-alkanoic or fatty acids (Table S6) because there were insufficient concentrations of n-alkanes in the OM from this section. For most compounds in both sections, irrespective of whether they are derived from terrestrial, aquatic or marine sources, the δ13C trends parallel that of the bulk OM to a greater or lesser extent (Figs. 9, 10; Tables S5, S6; Sect. S4).

Despite parallel trends, the magnitudes of the positive CIE differ considerably (Fig. 11). Of the compound-specific components, phytane exhibits the largest CIE (∼ 7 ‰), followed by pristane and the fatty acids (∼ 3 ‰–5 ‰), the LMW and MMW n-alkanes (2 ‰–3 ‰), and then the HMW n-alkanes (∼ 1 ‰). Unless noted otherwise, the CIEs referred to here are the difference between the mean of samples from OM-rich Waipawa organofacies and the mean of samples from the underlying facies (Tables S5, S6).

Variations in the magnitude of the CIEs amongst different compound classes are commonly due to mixing of OM sources, as discussed above for bulk fractions, or due to different isotopic sensitivities to environmental change in the source organisms (Pancost et al., 1999; Schouten et al., 2007). A strong correlation between pristane and phytane carbon isotope compositions and δ13COM (Table S5) suggests that they derive from a mixture of terrestrial and aquatic sources in the Taylor White section, with the terrestrial source dominant in OM-rich Waipawa organofacies. A positive CIE for pristane indicates that the primary terrestrial substrate may be enriched in 13C by ∼ 4 ‰ (Fig. 11c). The greater CIE for phytane suggests an additional unidentified source of enriched carbon.

The LMW and MMW n-alkanes and fatty acids, which are thought to be derived mainly from aquatic sources, exhibit positive CIEs of 2 ‰–5 ‰ (Fig. 11c, d). These shifts may reflect a change in substrate δ13C or secondary processes that enhance carbon isotope discrimination during photosynthesis such as a decline in atmospheric CO2 concentration or an increase in plant growth rates (Bidigare et al., 1997).

For the higher plant biomarkers, HMW n-alkanes at Taylor White exhibit a much lower CIE (∼ 1 ‰) than the HMW fatty acids at mid-Waipara (∼ 3 ‰). It seems that the small magnitude of the CIE for HMW n-alkanes at Taylor White may reflect some mixing of terrestrial OM sources. A combination of contemporaneous and older reworked sources was found to have dampened the signal from contemporaneous higher plants in a study of the Paleocene–Eocene transition in Tanzania (e.g. Carmichael et al., 2017). We find support for this possibility in additional compound-specific δ13C analyses undertaken on OM-rich Waipawa organofacies at Angora Road and Mead Stream (Fig. 12; Table S5). At these sites, Waipawa organofacies has similar δ13C values for phytane, pristane, LMW and MMW n-alkanes as at Taylor White. However, in contrast to the weakly enriched HMW n-alkane values at Taylor White, the HMW n-alkanes at Angora Stream and Mead Stream are as enriched as the shorter n-alkanes at these sites. This suggests that the relatively depleted HMW values at Taylor White may be due to mixing with an older source and the contemporaneous higher plant input is better represented by values of ∼ -27.5 ‰ recorded at Mead and Angora. This value indicates an excursion of ∼ 3 ‰, which is in line with the HMW fatty acids at mid-Waipara.

In summary, most of the organic compounds we have analysed are enriched in 13C in the OM-rich Waipawa organofacies relative to bounding facies. The evidence indicates a significant excursion of at least 2 ‰ for compounds derived from both aquatic and terrestrial OM. Crucially, we have demonstrated that the δ13C trends in the aquatic and terrestrial biomarkers parallel trends in bulk OM (Figs. 9, 10). This implies that the primary influence on δ13COM, the proportion of degraded woody plant matter, is modulated by the same carbon cycle changes that cause the variation in δ13C in aquatic and terrestrial biomarkers.

In order to establish how Waipawa deposition might be linked to global climate, we reviewed and revised the age control for Waipawa organofacies. Based on a combination of biostratigraphy (Schiøler et al., 2010; Crouch et al., 2014; Kulhanek et al., 2015) and limited magnetostratigraphy, Hollis et al. (2014) inferred that Waipawa organofacies deposition occurred over ∼ 700 kyr between ∼ 59.4 and ∼ 58.7 Ma (GTS2012, Gradstein et al., 2012). We have extended the bulk carbonate δ13C stratigraphy for the Paleogene section at Mead Stream (Hollis et al., 2005; Nicolo et al., 2007; Slotnick et al., 2012) to encompass the interval spanning the Waipawa organofacies (Fig. 13; Sect. S3). Mead Stream has the most complete known record of Waipawa organofacies in a distal setting (Fig. S5a), with a gradational base and sharp but potentially conformable top (Fig. S5). It also has a bulk carbonate δ13C record that parallels benthic and bulk carbonate records in deep-sea sediment cores (Hollis et al., 2005; Nicolo et al., 2007; Slotnick et al., 2012). This allows us to correlate the interval to high-resolution stable isotope records from North Pacific ODP Site 1209 and South Atlantic ODP Site 1262 (Westerhold et al., 2008, 2011, 2020; Littler et al., 2014; Barnet et al., 2019). We use the 2020 Geological Timescale for the Paleogene (Speijer et al., 2020), which incorporates the astronomical age control of Westerhold et al. (2008, 2011, 2017, 2020).

Five CIEs in the Paleocene bulk carbonate δ13C record form the basis of an age model that encompasses Waipawa organofacies deposition (Fig. 13): the early late Paleocene event (ELPE), 59.3 Ma; the unnamed “*” event of Littler et al. (2014) and Barnet et al. (2019), 58.15 Ma; the B2 CIE, 57.25 Ma; the C2 CIE, 56.85 Ma; and the PETM, 55.93 Ma (Fig. 13a). Although alternative correlations are possible, this age model is in best agreement with biostratigraphy for the section (Hollis et al., 2005) and age control from other sections (Hollis et al., 2014), does not require major changes in sediment accumulation rate within uniform lithofacies, and retains a consistent relationship with the record from ODP Site 1262 at which Mead Stream δ13C values tend to be slightly more depleted. This age model provides a revised age estimate for the main phase of Waipawa organofacies of 59.2–58.5 Ma, which is the same duration but slightly younger than previous estimates (Hollis et al., 2014) and consistent with a new age estimate of 57.5 ± 3.5 Ma derived from Re–Os geochronology (Rotich et al., 2020). In contrast to sections where sedimentation rates increase during Waipawa deposition, carbonate flux decreases in the pelagic setting and sedimentation rate decreases markedly (Hollis et al., 2014). A second 20 cm thick layer of Waipawa organofacies occurs 5 m above the main phase of deposition. Our age model suggests this layer is correlated with the “*” event. Both this CIE and the ELPE are possible hyperthermals but have inconsistent evidence for warming (Littler et al., 2014; Barnet et al., 2019). In contrast to the Eocene CIEs at Mead Stream, which are associated with marl-rich intervals in the micritic limestone succession (Hollis et al., 2005; Slotnick et al., 2012), no obvious changes in lithology are associated with the ELPE and the “*” CIE is centred on a thin siliceous Waipawa organofacies layer.

The main phase of Waipawa deposition spans the first 700 kyr of the Paleocene carbon isotope maximum (PCIM; Fig. 13b), a ∼ 1.9 Myr episode in which δ13C values for marine carbonate reach their Cenozoic maximum. The positive CIE associated with the PCIM is ∼ 1 ‰ in benthic foraminifera and in bulk carbonate and therefore accounts for only a small portion of the residual Waipawa organofacies CIE of 2 ‰–3 ‰. Waipawa deposition also occurs within a Paleogene maximum in marine carbonate δ18O, which extends from 59.6 to 58.2 Ma (Figs. 1, 13c). We refer to this interval here as the Paleocene oxygen isotope maximum (POIM). The positive shift in benthic δ18O signals cooling of bottom waters despite the interval including the two possible hyperthermals noted above: the ELPE and the “*” event, the latter coinciding with the termination of the POIM (Fig. 13a, b). In addition, Waipawa deposition coincides with a marked decrease in the coarse fraction in foraminiferal residues from ODP sites 1209 and 1262 (Fig. 1). This represents a marked decrease in the abundance of planktic foraminifera and is attributed to carbonate dissolution (Littler et al., 2014). Although short-lived dissolution episodes have been linked to the PETM and other early Eocene hyperthermals (Zachos et al., 2005; Alexander et al., 2015), the longer duration of this episode and its association with positive shifts in both δ13C and δ18O suggest a link to climatic cooling and carbon burial (Hilting et al., 2008).

Covariance between δ13COM and the TEX86 sea surface temperature (SST) proxy at mid-Waipara River and ODP Site 1172 (Fig. 10b, Sect. S4) indicates that Waipawa deposition is associated with regional cooling of coastal waters (Hollis et al., 2014; Bijl et al., 2021). Regional cooling on land is also indicated by temperature reconstructions based on pollen assemblages at Site 1172 (Contreras et al., 2014). Correlation with the POIM further suggests that the positive δ13C excursion in Waipawa organofacies (Fig. 13c–f) is linked to global cooling of ∼ 1 ∘C. Regional SST decreased by 4–6 ∘C, but this might reflect localized phenomena such as enhanced upwelling of Antarctic deep water (Hollis et al., 2014).

Carbon isotope shifts of ∼ 2 ‰–3 ‰ in marine and terrestrial biomarkers may be caused by a range of environmental factors, but a correlation with climatic cooling suggests that this positive CIE may be due to a decline in atmospheric CO2. A decline in CO2 will result in 13C enrichment in the biomass of algae (Freeman and Hayes, 1992) as well as in higher plants (Schubert and Jahren, 2012). Most other environmental factors, such as changes in lapse rate (Körner et al., 1988), would not affect terrestrial and marine δ13C to similar degrees.

We have used the relationship between atmospheric CO2 and C3 plant tissue δ13C values (Cui and Schubert, 2016, 2017, 2018; Schubert and Jahren, 2012, 2018) to estimate atmospheric CO2 concentrations prior to and during Waipawa deposition (Fig. 14). The change in δ13C (Δ13C) per part per million of CO2 follows a hyperbolic relationship (Schubert and Jahren, 2012) and is based on the model of carbon isotope fractionation in plants originally described by Farquhar et al. (1989). This proxy yields an estimate for CO2 that is based on the relative change in Δ13C between the time of interest (Δ13C(t)) and the Δ13C value at a chosen initial time (Δ13C(t=0)), which is designated as Δ(Δ13C) and expressed as Eq. (1): 1ΔΔ13C=ABCO2(t)+CA+BCO2(t)+C-ABCO2(t=0)+CA+BCO2(t=0)+C, where A, B and C are curve fitting parameters and solved for CO2 at any time t (CO2(t)) using Eq. (2) (Cui and Schubert, 2016): 2CO2(t)=ΔΔ13C⋅A2+ΔΔ13C⋅A⋅B⋅CO2(t=0)+2⋅ΔΔ13C⋅A⋅B⋅C+ΔΔ13C⋅B2⋅C⋅CO2(t=0)+ΔΔ13C⋅B2⋅C2+A2⋅B⋅CO2(t=0)A2⋅B-ΔΔ13C⋅A⋅B-ΔΔ13C⋅B2⋅CO2(t=0)-ΔΔ13C⋅B2⋅C. The combined uncertainty of parameters used to derived the estimate for atmospheric CO2 is relatively large and increases with increasing CO2 (Cui and Schubert, 2016, 2018).

As in Cui and Schubert (2018), we use the latest Paleocene (56.1–56.5 Ma, t=0) as the reference time and adopt the same parameters (Table 1) with some modifications. We exclude an unusually low estimate of 100 ppm for CO2 derived from paleosols by Sinha and Stott (1994), and we base our estimates for the δ13C of atmospheric CO2 (δ13CCO2; Fig. 14d) on the method described by Tipple et al. (2010) but recalculated using the smoothed LOESS benthic foraminiferal δ13C and δ18O compilation of Westerhold et al. (2020). For this calculation, we use the temperature equation of Kim and O'Neil (1997) rather than that of Erez and Luz (1983), which was developed for planktonic foraminifera and is not appropriate for benthic foraminiferal calcite (Hollis et al., 2019). We assume ice-free conditions for this calculation (i.e. δ18Oseawater​​​​​​​ = -1 ‰), while noting that the findings of this study imply the growth of ice sheets during Paleocene episodes. The three time slices used for our δ13CCO2 reconstructions are the latest Paleocene (pre-PETM) reference time slice (56–56.2 Ma), Waipawa organofacies (WOF; 58–59.2 Ma) and underlying organofacies (pre-WOF; 59.6–59.8 Ma) (Table 1, Fig. 14).

We have derived three estimates for the change in CO2 that can be linked to Waipawa deposition. These are based on estimated bulk terrestrial δ13C values as well as δ13C values for the higher plant biomarkers, odd-numbered HMW n-alkanes (C27–C33) and even-numbered HMW fatty acids (C26–C32). For our latest Paleocene (t = 0) δ13C values, we use the mean bulk and HMW n-alkane δ13C values from coal samples in the uppermost Paleocene in the Canterbury (New Zealand) Otaio River section (Inglis et al., 2021). In the absence of fatty-acid δ13C data for this section, we use the n-alkane data as reference for both n-alkanes and fatty-acid δ13C.

For HMW fatty acids in the mid-Waipara section, the carbon isotope excursion (CIE) from the mean value for underlying facies to the mean value for the main phase of Waipawa deposition is 2.2 ‰ (Table S6, mean values of -30.4 ‰ and -28.2 ‰). For HMW n-alkanes in the Taylor White section, we have argued that the HMW n-alkanes in the Waipawa facies have been affected by mixing. If we substitute values from the nearby Angora Road section (Table S5), we derive a CIE of 3.3 ‰ based on the average of two OM-rich Waipawa samples from Angora Road (value of -27.45 ‰) and a single sample from underlying Whangai facies in the Taylor White section (value of -30.7). Because we cannot be sure of the extent to which the bulk terrestrial δ13C values are affected by lignin alteration, we have adopted an intermediate value of 2.5 ‰ for the bulk organic CIE. We use the δ13C values from the density fractions from the Taylor White section (Sect. 5.2.1) to derive a value of -21 ‰ for terrestrial OM in underlying Whangai facies. A CIE of 2.5 ‰ implies a value of -18.5 ‰ for Waipawa organofacies. As the maximum value for Waipawa organofacies is -16.7 ‰, this suggests that lignin degradation may only account for ∼ 1.5 ‰ of the total excursion.

The three approaches result in significant differences in CO2 estimates, both for Waipawa facies and the underlying facies (Table 1). CO2 estimates range from 145 to 318 ppm for Waipawa organofacies and from 203 to 484 ppm for the underlying facies. This represents a 28 %–43 % decrease in CO2 during Waipawa deposition. This variation in values is to be expected given the many sources of uncertainty related to estimating the magnitudes of the CIEs for each parameter, variability within biomarkers and uncertainties in the calibration itself. On average, the different approaches indicate that a ∼ 35 % decrease in CO2 may be linked to Waipawa deposition. Temperature estimates derived from the benthic foraminiferal compilation indicate global temperature decreased by ∼ 1 ∘C from the pre-WOF to WOF time slices (Fig. 14c). An accompanying decrease of ∼ 35 % in CO2 implies a climate sensitivity of ∼ 3 (i.e. a 1 ∘C decrease in temperature for a 35 % decrease in CO2 equates to a decrease of 3 ∘C for a halving of CO2). As noted above, however, our temperature calculations assume ice-free conditions. If cooling was associated with ice growth, a portion of the positive shift in δ18O should be attributed to this increase in ice volume, which would lead to a smaller decrease in temperature and, therefore, lower climate sensitivity.

Our estimates for CO2 in the underlying facies are consistent with published estimates for CO2 in the Paleocene (Fig. 14e), with best fit shown by n-alkanes and fatty acids. This suggests that CO2 levels during Waipawa deposition were in the range of ∼ 300 ppm, i.e. below present-day levels and low enough for polar ice sheet growth, at least in the Southern Hemisphere. The low CO2 estimates derived from terrestrial OM suggest that we may have underestimated the degree to which lignin degradation has affected the δ13C of OM in both Waipawa organofacies and underlying sediments.

Waipawa organofacies is widely distributed in the southwest Pacific (Fig. 1), occurring in most of New Zealand's sedimentary basins as well as the East Tasman Plateau (Hollis et al., 2014). It is inferred to have been deposited at a range of paleo-depths, from inner shelf to middle slope (Moore, 1988; Schiøler et al., 2010; Naeher et al., 2019), and within a narrow time window of ≤ 1 Myr in the early late Paleocene (∼ 59 Ma). Previously, Waipawa organofacies was thought to have been deposited during a regression or lowstand following a base level fall (Schiøler et al., 2010; Hollis et al., 2014). However, benthic foraminiferal assemblages in the Taylor White section indicate a general deepening or transgressive trend from the underlying Whangai and into the overlying Wanstead Formation (Naeher et al., 2019). This suggests an alternative interpretation of the palynofacies assemblages within the Waipawa organofacies is required. If we examine how palynofacies assemblages vary in relation to proximity to paleo-shoreline for our studied sections (Fig. 15), we find that a conventional distribution is evident for the underlying facies, i.e. Whangai or Wickliffe formations (Fig. 15a). Terrestrial components (i.e. phytoclasts + terrestrial palynomorphs) tend to decrease, whereas marine elements (amorphous organic matter + marine palynomorphs) increase with water depth and hence with distance from shore. However, the relationship is reversed for Waipawa organofacies: terrestrial components increase, whereas marine components decrease with water depth and distance from shore (Fig. 15b). Degraded phytoclasts, in particular, exhibit a pronounced increase in abundance in the deeper and more distal sections. This distribution of terrestrial OM is supported by biomarkers, which show that terrestrial influence is strongest in sections with the thickest accumulations of Waipawa sediments (e.g. Taylor White and Orui-1A) and weakest in both more proximal (Te Hoe) and most distal (Mead, Ben More) settings (Fig. 16).

We interpret this to indicate that Waipawa organofacies is the result of a rapid influx of terrestrial OM into the marine environment. Corroborating evidence for a marked increase in terrestrial runoff at this time has recently been reported in a study of biomarkers and dinoflagellate paleoecology from ODP Site 1172 (Bijl et al., 2021). The foraminiferal data imply that this runoff event occurred while the East Coast Basin was progressively deepening as part of long-term passive margin subsidence throughout New Zealand (King et al., 1999). This discovery resolves a long-standing enigma in local geology. In the sedimentary basins to the east of New Zealand, a transition from siliciclastic to hemipelagic sedimentation occurs in the late Paleocene. In the central and northern East Coast Basin, the siliciclastic Whangai facies is overlain by the hemipelagic Wanstead facies (Field and Uruski, 1997). Both units are inferred to have been deposited at bathyal depths. If Waipawa organofacies is present, it occurs at the facies transition. It is difficult to develop a credible depositional model in which a nearshore Waipawa facies was sandwiched between two bathyal units, the overlying one being deeper bathyal than the underlying one.

Although the stratigraphic sections used in this study are primarily on the eastern margin of Paleocene Zealandia, there is evidence from the Taranaki Basin (Fig. 2) that intensified terrestrial runoff affected the entire landmass. Much of the basin was non-marine to shallow marine in the Paleocene, but in the offshore Reinga sub-basin to the north a 26 m thick interval of Waipawa organofacies is present in the Waka-Nui 1 exploration well (Stagpoole et al., 2009). This suggests that with further stratigraphic drilling in the offshore western basins, more records of Waipawa organofacies will likely be found.

When correlated to deep-sea benthic isotope records (Westerhold et al., 2011, 2020; Littler et al., 2014; Barnet et al., 2019), Waipawa organofacies deposition is found to coincide with a minimum in deep-sea temperatures and the onset of the ∼ 2 Myr long PCIM (Fig. 1). Several factors have been implicated in the long-term trends in Paleocene temperature and carbon cycling. Cooling from the late early to middle Paleocene (63–60 Ma) followed by warming in the late Paleocene (58–56 Ma) has been linked to global trends in volcanism (Westerhold et al., 2011), carbon cycling (Komar et al., 2013), continental rifting (Brune et al., 2017) and tectonism (Beck et al., 1995; Kurtz et al., 2003; Rotich et al., 2020). Kurtz et al. (2003) argued that the PCIM was caused by the accumulation of terrestrial carbon during tectonic uplift. They point to the vast coal deposits of the Powder River basin, which represent the swamps that replaced North America's epeiric seas during the Laramide orogeny. Our observation that the OM in the Waipawa organofacies is also terrestrial adds another, albeit offshore, sink for terrestrial organic carbon at this time. Hilting et al. (2008) also interpret the PCIM to be a time of enhanced terrestrial carbon burial and reduced CO2 levels. Terrestrial carbon burial is modelled to have reduced dissolved inorganic carbon (DIC) in the global ocean (Hilting et al., 2008). This is consistent with our correlation of Waipawa organofacies with an interval of carbonate dissolution during the initial part of the PCIM (Fig. 1). During the second part of the PCIM, carbonate accumulation recovered at the same time as deep-sea temperatures began to increase (Barnet et al., 2019), suggesting that a new source of carbon offset the effects of carbon burial, such as the CO2 outgassing from the second phase of North Atlantic Igneous Province (NAIP) volcanism (Westerhold et al., 2011) or oxidation of existing carbon reservoirs (Komar et al., 2013).

Paleocene sediments with organic δ13C values in the same range (-24 ‰ to -17 ‰) as those reported for Waipawa organofacies have been reported from non-marine Paleocene sections in China (Clyde et al., 2008) and Argentina (Hyland et al., 2015). In the middle-Paleocene Chijiang Basin section in China, a positive δ13COM excursion occurs from a background of values of -24 ‰ to -22 ‰ to -17 ‰ (Clyde et al., 2008). This demonstrates that similar processes may have affected terrestrial plant matter in regions beyond the southwest Pacific, i.e. CO2-controlled carbon fractionation with or without lignin degradation. However, magnetostratigraphy for this section indicates that the excursion predates the Waipawa event by ∼ 1 Myr. In the Salto Basin section in Argentina, a middle–late Paleocene section comprises two cycles in which δ13COM ranges from depleted values of -26 ‰ to -25 ‰ to enriched values of -21 ‰, again a shift similar to that observed in New Zealand. The enriched values are linked to proxies for lower temperature and reduced precipitation (Hyland et al., 2015). A 30 m thick interval with enriched δ13COM values at the base of the late Paleocene (lower Chron 26n) is very sparsely sampled and may correlate with the Waipawa event. It is sandwiched between two intervals with more depleted δ13COM values that are correlated with a two-phase ELPE but may represent the ELPE and “*” events.

Some have argued that the ELPE is a hyperthermal (Bernaola et al., 2007), but expression of the negative CIE is variable and evidence for warming is equivocal. Possible evidence is seen at ODP Site 1262 (Littler et al., 2014; Barnet et al., 2019), where individual samples record light δ18O values. However, these samples are anomalies against a background of relatively heavy δ18O values and may be more plausibly explained by downslope transport of individual benthic foraminifera. Moreover, a burrowed horizon near the base of the ELPE at Site 1262 (1262B-18H-4, 97–127 cm) suggests the presence of an unconformity and, therefore, an incomplete record.

The high-resolution records from sites 1209 and 1262 (Fig. 1) indicate that the ELPE marks a significant turning point in Paleocene climate and carbon cycling: the termination of a long-term (4 Myr) cooling trend followed by a prolonged period of carbon burial, i.e. the PCIM. Significantly, benthic foraminiferal δ18O and δ13C trends are coupled from 63 to 59 Ma, with positive shifts in both parameters suggesting that cooling was associated with declining atmospheric CO2 (Fig. 1). From 59 to 58 Ma, the trends are not coupled: δ13C continues to increase while δ18O either decreases or remains stable. From 58 Ma to at least the early Eocene, the records are once more coupled. This interval of uncoupled isotope records begins with the ELPE and ends with the “*” event. Barnet et al. (2019) show that the interval also contrasts with intervals below and above by having much less coherent eccentricity phasing, suggesting the influence of non-orbital climate drivers, such as the tectonic events noted above. Carbon cycle modelling also indicates that this interval is followed by a shift to a net decrease in organic carbon burial, with oxidation of carbon reservoirs driving the subsequent warming trend that culminates in the early Eocene climatic optimum (Komar et al., 2013). This shift may be linked to changes in deep-water circulation. Neodymium isotopes point to intensified deep-water exchange between the North and South Atlantic, probably due to the deepening of the central Atlantic Rio Grande rise (Batenburg et al., 2018).

In summary, there is wide-ranging evidence that the interval of Waipawa organofacies deposition is linked to a significant turning point in Paleogene climate and carbon cycling, transitioning from the cooler climate conditions and relatively high rates of organic carbon burial of the middle Paleocene to the warming climate and lower rates of organic carbon burial of the late Paleocene and early Eocene.