Elevated Post K-Pg Export Productivity in the Gulf of Mexico and 1 Caribbean 2

16 The global heterogeneity in export productivity after the Cretaceous-Paleogene (K-Pg) mass 17 extinction is well documented, with some sites showing no change on geologic timescales, some 18 demonstrating sustained decline, and a few showing a somewhat surprising increase. However, 19 observational data come from sites so widespread that a key outstanding question is the geographic scale 20 of changes in export productivity, and whether similar environments (e.g., open ocean gyres) responded 21 similarly or whether heterogeneity is unrelated to environment. To address this, we developed three new 22 Ba/Ti export productivity records from sites in the Gulf of Mexico and Caribbean which, combined with

published data from a fourth site in the Chicxulub Crater itself, allow us to reconstruct regional changes in 24 post K-Pg export productivity for the first time. We find that, on a regional scale, export productivity 25 change was homogenous, with all four sites showing a ~300 kyr period of elevated export production just 26 after the boundary, followed by a longer period of decline. Interestingly, this interval of elevated export 27 production appears to coincide with the post K-Pg global micrite layer, which is thought to at least 28 partially have been produced by blooms of carbonate-producing cyanobacteria and other 29 picophytoplankton. Global comparison of sites shows that elevated export productivity appears to have 30 been most common in oligotrophic gyres, which suggests that changing plankton ecology evidenced by 31 the micrite layer altered the biological pump, leading to a temporary increase in export production in 32 these settings. 33

Introduction 48
The end Cretaceous mass extinction was associated with a severe disruption of marine 49 productivity (Hsü and Mackenzie, 1985;Zachos et al., 1989;D'Hondt et al., 1998;Coxall et al., 2006;50 Birch et al., 2016). A reduction in sunlight received at Earth's surface caused by dust, soot, and sulfate 51 aerosols ejected by the Chicxulub impact resulted in a reduction in photosynthesis which is thought to 52 have led to the collapse of marine food webs (Alvarez et al., 1980 removing the proximal external stress on marine primary producers and clearing the way for the recovery 56 of primary production. How, exactly, marine productivity recovered has been a focus of K-Pg boundary 57 research for decades; the K-Pg mass extinction represents a geologically unique disruption of marine 58 ecosystems, perhaps the only major event in Earth history which happened faster than modern climate 59 change and environmental disruption. Modern oceans are likely on the verge of a major reorganization of 60 dominant plankton types due to warming, acidification, and changes in circulation and ventilation patterns 61 (e.g., Barton et al., 2016;Jonkers et al., 2019), and primary production is expected to decline 20% due to 62 warming (Moore et al., 2018). The earliest Paleocene provides a window into understanding how such 63 ecological changes may impact food webs and marine carbon burial. 64 Of course, we can't observe ancient primary production in the euphotic zone directly, so most 65 work on the collapse and recovery of productivity after the K-Pg boundary has focused on sedimentary 66 records of export production (the transfer of particulate organic matter from the euphotic zone to the deep 67 sea; e.g., Passow and Carlson, 2012)). The movement of particulate organic matter (POM) from the 68 euphotic zone to the seafloor is complicated and can be divided into a series of steps, all of which are 69 influenced by different processes. Most net primary production (NPP) occurs in the euphotic zone 70 (dependent on sunlight penetration but typically defined as 0-100 or 200 m water depth; Passow and 71 Carlson, 2012), and most POM is remineralized in these near surface waters. The precise amount varies 72 by region and season, but typically ~ 90% of NPP is consumed and recycled before it can sink out of the 73 7 elemental Ba data need to be normalized against a terrigenous element like titanium or aluminum to 149 control for any possible detrital barium component (e.g., Dymond et al., 1992 correlated directly to the export flux of POM, but the ocean is undersaturated in barite, which means that 160 70% of particulate barite (and more in anoxic regions) dissolves in the water column and the upper few 161 cm of the sediments before it is buried (Carter et al., 2020). This means that the replacement of one 162 watermass with another of a different Ba 2+ saturation state could lead to a change in barite accumulation 163 which could be misinterpreted as a change in export flux (e.g., Carter et al., 2020). 164 These caveats make it difficult to directly extrapolate from Ba flux to absolute values of export 165 flux in mass of organic carbon per unit time, particularly all the way back in the Paleocene, but if major 166 variables (terrigenous flux, water mass changes) are controlled for, then marine barite can provide 167 important insights to changes in export flux. Hull and Norris (2011) used XRF-derived Ba/Ti and Ba/Fe 168 ratios from five K-Pg boundary sites to bolster the export productivity record of benthic foraminifera, and 169 demonstrated that changes in export production across the boundary were indeed geographically 170 heterogeneous, with some sites showing an increase in export production after the boundary. 171 Understanding geographic heterogeneity in export production is necessary to understand the 172 overall recovery of marine primary producers after the K-Pg boundary. In particular, the calcareous have adaptations which indicate a mixotrophic lifestyle (i.e., they supplemented photosynthesis by 180 ingesting small prey like bacteria); later incoming taxa lack these adaptations, indicating changing trophic 181 conditions (specifically the under exploitation of small prey species following the extinction of many 182 heterotrophic plankton) may have played a role in nannoplankton recovery (Gibbs et al., 2020). The 183 timing of these acme events is geographically variable, and at sites with elevated export productivity after 184 the K-Pg (Shatsky Rise and Chicxulub Crater), it is coincident with an observed decline in export 185 production (Jones et al., 2019). In the ocean today, eutrophic waters tend to be dominated by a few taxa 186 best suited to take advantage of widely available food, while oligotrophic waters tend to have much 187 higher diversity with greater degrees of specialization (e.g., Hallock, 1987 hypothesized that the recovery of primary producer assemblages (and by extension the ecosystems which 189 they supported) after the K-Pg is similarly linked to nutrient state controlled by the recovery of the 190 biological pump, but the linkages are not well understood and a better picture of export productivity 191 trends is a necessary first step. 192 Unfortunately, geographic trends in early Paleocene export productivity are still poorly known. regional-scale study (~1700 km) of export productivity after the K-Pg. This region was modelled to have 210 been characterized by low export production in the latest Cretaceous (Henehan et al., 2019) and thus may 211 be predicted to have exhibited increased export production after the boundary. We found that earliest 212 Danian export productivity is elevated at all Gulf of Mexico and Caribbean sites and that an initial 213 reduction in export production occurs ~ 300 kyr after the boundary at all sites, indicating that export 214 productivity trends were homogeneous at a regional scale. 215  Pindell and Barrett (1990) and Snedden et al. (2021). Black indicates land and grey indicates continental platforms.

Study Sites 216
We looked at three scientific ocean drilling sites in the greater Caribbean region with a well-217 preserved K-Pg boundary interval and compared them to published XRF data from IODP Site M0077 in 218 the Chicxulub Crater (Figure 1). An additional site, DSDP Site 536, below the Campeche Escarpment in 219 the southeastern Gulf of Mexico (Buffler et al., 1984), was considered but rejected because a preliminary 220 examination of planktic foraminifera in the nominally lowermost Paleocene cores found a mix of 221 biozones ranging from the Cretaceous to the late Paleocene, indicating significant reworking and/or 222 drilling disturbance, suggesting that XRF data would be untrustworthy. All four sites appear to have been  Site 1001 was drilled in 1995-6 during ODP Leg 165, and is located on the Hess Escarpment on 239 the Nicaragua Rise ( Figure 1). Shipboard biostratigraphy placed the K-Pg boundary between Core 240 1001A-38R-CC and 1001A-39R-1 ( Figure 2B). Unlike the thick K-Pg boundary deposits in the Gulf of 241 Mexico, here the whole interval is just a few cm thick. Maastrichtian limestone is overlain by a 1 cm thick 242 dark greenish gray clay, which is in turn overlain by a 3.5 cm bluish gray claystone containing 1 mm 243 scale dark green spheroids interpreted to be tektites (Sigurdsson et al., 1997). This tektite layer is turn 244 were then covered with 4 µm thick Ultralene film to prevent sediment from sticking to the scanner. 294 Cores were scanned on an Avaatech XRF Core Scanner at two excitation conditions focused on 295 different element groups. The first scan was at 10kVp with no filter to analyze major and minor elements 296 (Al, Si, K, Ca, Ti, Mn, Fe, Cr, P, S, and Mg) and the second was at 50kVp with a Cu filter to analyze 297 heavier trace elements (Sr, Rb, Zr, and Ba). Scan resolution was set depending on relative distance above 298 the K-Pg boundary, based on low resolution shipboard biostratigraphy. Core sections within Zone Pα or 299 the lower part of Zone P1a (very roughly, within ~ 500 kyr after the boundary) were scanned at 1 or 2 cm 300 steps, and sections below the boundary and >500 kyr after the boundary were scanned at 5 cm steps. 301 Some steps were skipped or moved based on visual examination of the core before scanning (e.g., to 302 avoid cracks or uneven surfaces). Laboratory standards were run at the beginning and end of each day to 303 monitor instrumental performance. 304 Raw spectral data were processed into peak areas in the lab and exported as count data using the 305 software program bAxil. Quality control of processed data was carried out using the following 306 parameters: 1) throughput (samples with values <150,000 cps, which indicates a gap between the sensor 307 and the core, were removed); 2) Argon peak (samples with positive Ar values, indicating that the sensor 308 was measuring ambient air, were removed); and 3) standard deviation (samples with elemental peaks of 309 Ba or Ti within 2 standard deviations of zero were removed). 310 To improve the age model for this study we analyzed planktic foraminifera from Site 95 at a 311 resolution of up to 5 cm. Lightly lithified samples were gently broken into cm-sized pieces using a mortar

Site 999 387
Site 999 is the southernmost site, and the most distal from the Chicxulub impact crater. Ba/Ti 388 values are very low directly above the boundary layer, quickly increasing through the lower part of Zone 389 Pα. Higher values after this brief recovery interval do no exceed the Ba/Ti values observed in the 390 uppermost Cretaceous, but they are much higher than subsequent Paleocene values (with the exception of 391 a brief peak around 2 myr after the K-Pg boundary). Once again, values decreased sharply approximately 392 320 kyr post-impact followed by gradually declining values. The very poor quality of the biostratigraphy 393 in this core (the Pα/P1a zonal boundary marker is the only reliable datum) makes it difficult to determine 394 the timing of this decline and whether the increase observed below the coring gap occurred just 600 kyr 395 after the boundary or much later. While we do not place much confidence in the ages above this level, we 396 are confident in age control in the key interval above the boundary, specifically the highest occurrence of 397 P. eugubina (Pα/P1a zonal boundary). 398

Regional Homogeneity of Post K-Pg Export Production 400
The most striking feature of the four export productivity records presented here, and the key 401  There are several possible mechanisms which could drive an increase in marine barite production 421 while export production is kept steady. Different groups of plankton incorporate different amounts of Ba 422 into their biomass. For example, coccolithophores have less Ba in their cells than diatoms, which in turn 423 have less Ba in their cells than chrysophytes (gold algae), which have less Ba than chlorophytes (green 424 algae) (e.g., Paytan and Griffith, 2007). Calcareous nannoplankton suffered a severe extinction at the K-425 Pg, and if they were briefly replaced in oligotrophic gyres by any of these other groups, the biogenic Ba broken apart more POM, slowed sinking and increased the time it was exposed to remineralization. Many 436 of these change (mesopelagic temperature increase, shifts in the grazer community or toward 437 phytoplankton with higher Ba abundance in their cells) are impossible to test with existing 438 paleoceanographic tools. What we can do, though, is look to other parts of the biological pump and see if 439 they indicate whether the observed increase in Ba/Ti was indeed related to export productivity. 440 Benthic foraminiferal accumulation rate and assemblages provide additional export productivity 441 information at these sites. Benthic foraminifera, which are responsive to the amount and quality of 442 organic matter that reaches the seafloor, record a different part of the biological pump than biogenic 443 barium, which is formed during the remineralization of organic matter at mesopelagic depths. foraminifer data, they both indicate increased transport of POM out of the euphotic zone and to the 457 seafloor; we therefore interpret the Ba/Ti data at all our sites as primarily recording an increase in export 458

productivity. 459
In the Gulf of Mexico and Caribbean, the interval of highest export production ends right around 460 the Pα/P1a zonal boundary at each site, roughly 300 kyr after the K-Pg boundary, followed by a general 461 decline over the next million years or so. The precise features vary from site to site; notably, the 462 prominent early peak observed in the Gulf of Mexico (Sites M0077 and 95) is absent in the Caribbean 463 cores (Sites 999 and 1001). Likewise, Site 999 records very low values immediately above the K-Pg 464 boundary followed by a rapid recovery that is not evident at any of the other sites. Finally, the timing of 465 the sharp decline of these high productivity intervals varies by a few tens of kyrs between sites. Because 466 the age models are based on biostratigraphy or paleomagnetic reversals, with no higher resolution 467 techniques like orbital chronology, it is impossible to say whether these differences are real or merely 468 artifacts of the limits of the age models. These are superficial differences, though, and a clear overall 469 trend exists that export productivity was elevated across Gulf of Mexico and Caribbean (a distance of 470 ~1700 km) for ~300 kyr after the K-Pg mass extinction, and began to decline thereafter. 471

23
The observed homogeneity in regional export productivity in the earliest Paleocene provides 472 important context for previous observations of global-scale heterogeneity determined with the Ba proxy. 473 Previous work had shown major differences in the amount of organic matter remineralized in the 474 mesopelagic zone between ocean basins, with an increase in export production in the middle of the North 475 Pacific, a decline in the western North Atlantic, western South Atlantic, and Southern Ocean, and no 476 change in the eastern South Atlantic (Hull and Norris, 2011). Those sites are widely separated and 477 represent different oceanographic environments (oligotrophic gyres, western boundary currents, eastern 478 boundary currents). With only one site in each region, it is hard to know whether these observations are 479 indicative of regional trends or more limited, local change. With the discovery that open ocean sites 480 within the Gulf of Mexico/Caribbean all exhibit the same trends (and, interestingly, the local change in 481 benthic foraminiferal diversity in nearshore environments Gulf of Mexico is also lower than many other 482 sites; Alegret et al., 2022), we can be more confident that previously observed regional differences are 483 real, and therefore conclude that oligotrophic open ocean sites were prone to increased export production 484 immediately after the K-Pg boundary, as suggested by Henehan et al. (2019). But what was the driver for 485 this increased export production? 486

Drivers of Post-Extinction Export Productivity 487
In the modern ocean, oligotrophic gyres are typically dominated (in terms of biomass) by 488 picophytoplankton (0.2-2.0 μm in size) like cyanobacteria and algae, but larger nano and micro 489 phytoplankton (2-20 μm and >20 μm, respectively), though less numerous, account for the majority of 490 productivity measured in incubation experiments (e.g., Marañón et al., 2003). Because picophytoplankton 491 have no physical fossil record, we cannot say for sure whether this was the case at the end of the 492 Cretaceous, but this seems like a safe assumption. nutrients could actually lead to an increase in primary and/or export productivity. 501 But how would NPP dominated by picophytoplankton lead to increased export production? After 502 all, if export increased then there would be a mechanism to remove nutrients from the euphotic zone and 503 NPP would necessarily decrease. Yet our work and that of others has found that high export production 504 was maintained in typically oligotrophic regions for 100-300 kyr Thomas, 2005, 2009; An alternate explanation could be the occurrence of blooms of specific groups of phytoplankton 518 with barium-rich cells or which favor barite formation. For example, in the modern ocean Phaetocystis is 519 a common haptophyte which secretes extracellular polymers which form aggregates that speed sinking 520 and enhance export production (e.g., Verity et al., 2007). These polymers may also play a key role in 25 marine barite formation as nucleation sites (Martinez-Ruiz et al., 2020). Acantharians have barium-rich 522 skeletons and are known to form blooms in oligotrophic regions (e.g., Decelle et al., 2012) but, like the 523 other groups, do not typically fossilize. Blooms of plankton like these may serve to increase export to the 524 seafloor and also increase marine barite production without necessarily relying on a stronger microbial 525 loop in the euphotic zone. While we currently lack direct evidence of blooms of non-fossilizing 526 phytoplankton like these groups, more work is required to provide a clear answer to this question. But we 527 can see some evidence for ecosystem changes associated with increased export productivity after the K-528 Pg. 529

Evidence of Ecosystem Changes 530
The second-most striking feature of our data is that at two of the sites studied (Sites 95 and 999) 531 the interval of high export productivity ~300 kyr after the boundary coincides almost exactly with well-532 so a precise tie between the decline in export productivity and the end of micrite deposition is impossible 561 to make. 562 Although various types of "ballast," including calcite plankton shells, have been thought to 563 influence export production in the modern ocean (Armstrong et al., 2001;Francois et al., 2002), it does 564 not seem likely that micrite itself, or more specifically the cyanobacteria that produced it, is the cause of 565 increased export production in the earliest Paleocene. Micrite is abundant at many sites which did not 566 experience elevated export production after the K-Pg. For example, Blake Nose Site 1049, which 567 experienced either a decline or no change in export production after the boundary (Alegret and Thomas, Our XRF-derived Ba/Ti export productivity proxy data from the Gulf of Mexico and Caribbean 596 show a post K-Pg peak in export productivity across the region, with an interval of high values lasting for 597 ~ 300 kyr after the boundary, then declining values for another ~ 700 kyr. This is a major improvement on 598 previous compilations of earliest Paleocene export productivity, which showed that post-extinction 599 changes in export production were globally heterogeneous on an ocean basin scale. Our results show that 600 broad regions followed similar trends. In particular, we find that most elevated export production in the 601 earliest Danian occurred tropical open ocean sites (Shatsky Rise, Hess Rise, and our Caribbean/Gulf of 602 Mexico sites) which were oligotrophic at the end of the Cretaceous (Henehan et al., 2019). 603 At sites with elevated export production and at which preservation makes such observations 604 possible, the post K-Pg global micrite layer corresponds to the interval of elevated export production. We 605 interpret this as evidence that the dominance of picophytoplankton like cyanobacteria and chlorophyte 606 algae, which appear to have been responsible for the micrite deposition (Bralower et al., 2020), altered the 607 dynamics of the biological pump to increase recycling of organic matter in the euphotic zone. Enhanced 608 recycling of organic matter left only refractory material, which is more difficult to recycle, to be exported 609 from the euphotic zone. Because it is refractory, this organic matter would have been more likely to sink 610 through the water column than more labile material exported under normal conditions. In typically 611 oligotrophic environments, this slight increase in efficiency of the biologic pump could have resulted in 612 overall higher export production; as larger phytoplankton recovered and more labile organic matter was 613 exported and grazed, enhanced export production would have subsided. 614 More datasets from a wider range of latitudes and ocean basins are needed to build a more 615 complete picture of post K-Pg export production to more fully understand how the marine biosphere 616 recovered from the most recent major mass extinction. 617

Open Research 618
XRF core scan data and age models are archived at the NCEI Paleoclimate Database (Lowery, 2021). 619 29 620

Acknowledgements 621
We are grateful for insightful reviews by Ellen Thomas and an anonymous reviewer, both of which 622 substantially improved this manuscript. We are also grateful to Brian LeVay and Mackenzie Schoemann 623 of the IODP Gulf Coast Repository (GCR) at Texas A&M University for their assistance with XRF core 624 scanning, the staff of the GCR for sending samples from Sites 95 and 536 for biostratigraphic analysis, 625 and Vinny Percuoco at the GCR for providing high resolution photograph for the K-Pg boundary in Hole 626 1001B. We are also grateful to Ryan Weber and Calvin Gordon of PaleoData, Inc., for their assistance 627 preparing samples from Site 999 and 1001 for biostratigraphic analysis. 628