Terrestrial environmental change across the onset of the PETM and 2 the associated impact on biomarker proxies : a cautionary tale 3 4

25 The Paleocene-Eocene Thermal Maximum (PETM; ~ 56 million years ago) is the most 26 severe carbon cycle perturbation event of the Cenozoic. Although the PETM is 27 associated with warming in both the surface (up to 8°C) and deep ocean (up to 5°C), 28 there are relatively few terrestrial temperature estimates from the onset of this interval. 29 The associated response of the hydrological cycle during the PETM is also poorly 30 constrained. Here, we use biomarker proxies (informed by models) to reconstruct 31 temperature and hydrological change within the Cobham Lignite (UK) during the latest 32 Paleocene and early PETM. Previous work at this site indicates warm terrestrial 33 temperatures during the very latest Paleocene (ca. 22-26°C). However, biomarker 34 temperature proxies imply cooling during the onset of the PETM (ca. 5-11°C cooling), 35 inconsistent with other local, regional and global evidence. This coincides with an 36 increase in pH (ca. 2 pH units with pH values > 7), enhanced waterlogging, a major 37 reduction in fires and the development of areas of open water within a peatland 38 environment. This profound change in hydrology and environment evidently biases 39 biomarker temperature proxies, including the branched GDGT paleothermometer. 40 This serves as a cautionary tale on the danger of attempting to interpret biomarker 41 proxy records without a wider understanding of their environmental context. 42


Introduction 49
The Paleocene-Eocene Thermal Maximum (PETM; ca. 56 million years ago) is a rapid 50 global warming event associated with the release of 13 C-depleted carbon into the 51 ocean-atmosphere system. During the PETM, the deep ocean warmed by ~5°C 52 Continental temperatures are also important because they exert a first-order 65 control upon the hydrological cycle. During the PETM, proxies and models indicate 66 that the hydrological cycle exhibits a globally 'wet-wetter, dry-drier' style response 67 (Carmichael et al., 2017). However, there is significant regional and temporal 68 variability in both proxy and model data. For example, high-latitude and coastal 69 settings are generally characterised by stable and/or increasing rainfall, with proxy 70 evidence for both enhanced terrigenous sediment flux to marginal marine sediments 71 (John et al., 2008) and enhanced chemical weathering (Dickson et al., 2015;Ravizza 72 et al., 2001). In contrast, mid-to-low latitude and continental interior settings are typically characterised by decreasing rainfall but an increase in extreme precipitation 74 rates (Carmichael et al., 2018;Handley et al., 2012;Schmitz and Pujalte, 2007). 75 Perturbations to the hydrological cycle also impacted vegetation patterns (Collinson et 76 al., 2009;Jaramillo et al., 2010) and various biogeochemical cycles (e.g. methane 77 cycling; Pancost et al., 2007), and may have played an important role in maintaining 78 the warmth of the PETM (Zachos et al., 2008b) and in the subsequent recovery phase 79 (Gutjahr et al., 2017). 80 To reconstruct temperature and hydrological change during the PETM, we 81 investigate the biomarker distributions within an immature lignite seam from Cobham, 82 Kent, UK (∼48°N palaeolatitude). The Cobham Lignite Bed is inferred to represent an 83 ancient continental mire system and is characterised by a negative carbon isotope 84 excursion characteristic at the PETM onset (Collinson et   (1) MBT'5ME = (Ia + Ib + Ic)/(Ia + Ib + Ic + IIa + IIb + IIc + IIIa) 160 For application to peats and lignites MBT'5ME is translated to MAAT using the peat-161 specific calibration (Naafs et al., 2017): 162 (2) MAATpeat = 52.18 * MBT'5ME -23.05 (n = 94, r 2 = 0.76; RMSE = 4.7°C) 163

Roman numerals refer to individual GDGT structures shown in the Supplementary 164
Information ( Figure S1). In brief, I, II and III represent the tetra-, penta-and 165 hexamethylated components, respectively, and a, b and c represent the brGDGTs 166 bearing 0, 1 or 2 cyclopentane moieties. Penta-and hexamethylated brGDGTs can be methylated at the C-5 position or C-6 position on the alkyl chain. The latter are 168 indicated by an apostrophe (e.g. IIa'see equation (7)). Note that samples from the 169 lower laminated lignite (i.e. pre-PETM; n = 7) were previously analysed for branched 170 GDGTs. For more details, see Naafs et al., (2018b). 171 Recent work has demonstrated that the distribution of bacterial-derived branched 172 glycerol monoalkyl glycerol tetraethers (brGMGTs) can also be influenced by MAAT,173 with the degree of methylation decreasing as temperature increases (Naafs et al.,174 2018a). This is captured in the H-MBTacyclic index: 175

211
We also generate new temperature and precipitation estimates using a revised 212 version of HadCM3L, HadCM3L-I2 (Table 1)  have indicated that vegetation change is less of a concern than originally inferred by 346  (Fig. 3). The occurrence of isoGDGT-5 (> 1%; Fig. 2d), the absence of 6-371 methyl brGDGTs (IR6ME < 0.01; Fig. 3c) and the dominance of the C31 αβ hopane 372 provides additional evidence for acidic conditions within the lower laminated lignite. 373 We observe a remarkable increase in hopanoid-and brGDGT-derived pH estimates

Conclusions 446
Here we have reconstructed terrestrial paleoenvironmental change within the Cobham 447 Lignite Bed, which spans the very latest Paleocene, onset and early part of the PETM. 448 consistent with model simulations. However, inconsistent with local, regional and 450 global evidence, the biomarker proxies seem to indicate significant cooling during the 451 onset and early PETM (ca. 5 to 11°C). We attributed this to enhanced waterlogging 452 and the development of a persistent peatland with areas of open water, biasing the 453 brGDGT paleothermometer. This study implies the need for care when applying 454 biomarker-based temperature proxies in transitional terrestrial environments (e.g. mire 455 settings). It also serves as a cautionary tale on the danger of attempting to interpret 456 proxy records without a wider understanding of the environmental context, especially 457 the pH and the hydrology.

Data availability 471
Data can be accessed via the online supporting information, via http://www. 472 pangaea.de/, or from the author (email: gordon.inglis@bristol.ac.uk). 473