The burial of marine planktic organism remains on the seafloor represents the capture and storage of carbon from the atmosphere, and a significant component of the long-term carbon cycle (Ridgwell and Hargreaves, 2007). A recent study indicates that carbon sequestered through pelagic sediments has become the dominant source of the deep carbon cycle since the early Cenozoic, peaking at ∼277 Mt C/yr of carbonate carbon flux at present, out of a total deep carbon input of 311 Mt C/yr across the oceanic plate system (Müller et al., 2022). These computations are based on the global carbonate compensation depth (CCD) estimates of Boss and Wilkinson (1991). The CCD is defined as the water depth at which the supply of carbonate is balanced by its dissolution, leading to the absence of carbonate-bearing sediments below the CCD (Lyle, 2003; Rea and Lyle, 2005). The amount of carbonate deposited in deep-sea sediments over time is critically linked to CCD variations over geological time because the position of the CCD determines the area of seafloor available for carbonate sedimentation (Dutkiewicz et al., 2018). Factors affecting carbonate burial, such as ocean saturation state and alkalinity feedback, the biological pump, and atmospheric CO₂, exhibit strong links to CCD shifts (Xiao et al., 2024). All these factors result in significant variability of the CCD between ocean basins (e.g., Campbell et al., 2018; van Andel, 1975; Xiao et al., 2024). For example, Campbell et al.’s (2018) results indicate that the CCD has varied by ∼300 m in the equatorial Pacific Ocean since the Pliocene, whereas the equatorial Indian Ocean has experienced fluctuations of ∼1 km during the same period. Recent reconstructions of regional CCDs within the Atlantic Ocean demonstrate that the CCD can differ contemporaneously by as much as 1.2 km even within a single ocean basin due to regional variabilities in productivity and ocean chemistry (Dutkiewicz and Müller, 2022). This highlights the importance of regional CCD reconstructions for robust calculations of deep-sea carbonate storage across the global ocean, which provide significantly more accurate inputs for solid Earth carbon cycle modeling (e.g., Müller et al., 2022). To date, such detailed computations have only been carried out for the Atlantic Ocean (Dutkiewicz and Müller, 2022).

The Pacific occupies a vast area of the Earth’s total surface covered by oceans, and plays a critical role in the recycling of carbon via its associated network of subduction zones (Müller et al., 2022). Yet, despite its latitudinally-variable accumulation of carbonate (Lyle, 2003) related to productivity (Pennington et al., 2006), upwelling (Gray et al., 2018), oceanic gateway configurations (Straume et al., 2020), geochemistry (Ma et al., 2018) and deep water circulation (Yin et al., 2022), our understanding of regional long-term variations of the CCD and carbonate flux throughout the Pacific is incomplete. The eastern equatorial Pacific is the only region that has been extensively studied (e.g., Campbell et al., 2018; Komar and Zeebe, 2021; Leon-Rodriguez and Dickens, 2010; Pälike et al., 2012; Rea and Lyle, 2005; van Andel, 1975), exhibiting slightly divergent CCD reconstructions for this region spanning the Cenozoic. Limited low-resolution CCD reconstructions based on a small number of deep-sea drill sites also exist for the northwestern Pacific (Rea et al., 1995) and the southeastern Pacific (Rea and Leinen, 1986). However, in the absence of regional CCDs, the high-resolution CCD of Pälike et al. (2012) in an unusually high-productivity region of the eastern equatorial is usually taken to represent the entire Pacific. Here, we present new CCD reconstructions for six regions of the Pacific spanning the Neogene and examine the influence of climate and oceanography changes on carbonate burial and CCD variability over time. The new CCD models will improve constraints on deep carbon cycling computation across the Pacific Ocean.

The Pacific CCD of van Andel et al. (1975) over the Cenozoic generally tracks the global CCD variations (e.g., Boss and Wilkinson, 1991; Delaney and Boyle, 1988). Van Andel et al. (1975) described the CCD in the equatorial Pacific as distinct from the entire Pacific CCD since 50 Ma, primarily in terms of the magnitude of fluctuations (∼500 m deeper), which aligns with the overall trend of the equatorial Pacific CCD of Berger (1973). However, our results show significant regional variability in the CCD fluctuations across the entire Pacific Ocean, highlighting the shortcomings of a single representative CCD over the entire Pacific (Figures 3, 4, 6). However, given that the eastern equatorial CCD is the most detailed and intensely studied CCD in the Pacific (Campbell et al., 2018; Lyle, 2003; Pälike et al., 2012), and is also well-defined in our reconstructions, we explore regional CCD changes in the Pacific in relation to the CCD model of the eastern equatorial area.

Our CCD reconstructions provide new insights into regional CCD variations across the Pacific since the early Miocene, particularly in the equatorial region that includes the first CCD model for the western tropical Pacific. Distinct CCD fluctuations in the western and eastern equatorial regions of the Pacific (Figures 4A,B) highlight the importance of longitudinal CCD variability beyond the previously discussed latitudinal gradient (Lyle, 2003). Previous studies (Campbell et al., 2018; Lyle, 2003; Pälike et al., 2012) have documented the eastern equatorial Pacific CCD evolution in detail. However, our complementary western tropical CCD model represents a new CCD reconstruction in this region (Figure 4B), revealing two significant (∼1 km) CCD deepening pulses from the early Oligocene to the early Miocene not seen in the eastern equatorial region. The first deepening, from ∼3.2 km at 30.5 Ma to ∼4 km at 29 Ma, remains enigmatic. It cannot be linked to regional gateway modifications due to a lack of any significant changes in tropical Pacific gateways before the Miocene (Lyle et al., 2008), nor to the early Oligocene East Antarctic glaciation, as this continental glaciation did not lead to deep ocean circulation changes or associated cooling (Lear et al., 2000; Mackensen, 2004). Further constraints from additional drill sites with reliable age models covering the earliest Oligocene, which are currently not available, are essential to determine whether this shift represents a regional response to primary carbonate productivity (e.g., Lyle et al., 2008; Moore Jr et al., 2004; Van Andel and Moore Jr, 1974).

The second CCD deepening in this region, from ∼3.2 km at 23.5 Ma to ∼4.1 km at 20.5 Ma following a gradual rise, may be associated with the significant growth of West Antarctic Ice Sheet (WAIS) at this time. A recent study by Klages et al. (2024) based on Earth system and ice sheet models, utilizing multiproxy data from both terrestrial and marine microfossils from West Antarctica’s Pacific margin, revealed that the West Antarctic ice sheet initially expanded into the marine realm at ∼26 Ma, lagging the East Antarctic glaciation by ∼7–8 Ma. A delayed response to this late Oligocene growth in Antarctic ice sheet has also been observed in ocean cooling of ∼1°C after ∼24 Ma based on geochemical proxies (Billups and Schrag, 2002). Thus, the western tropical CCD deepening after ∼24 Ma may be ascribed to the ocean’s delayed cooling in response to the late Oligocene WAIS, coinciding with a notable decrease in atmospheric CO₂ during the earliest Miocene (Hönisch et al., 2023). Our results suggest that the early Miocene Antarctic Bottom Water (AABW) formation (Woodruff and Savin, 1989), following the expansion of the WAIS and its northward flow along the western South Pacific, may have impacted the western tropical Pacific during the earliest Miocene. This glacial cooling may be incorporated with low sea surface temperature (SST) affected by the distributed Western Pacific Warm Pool (WPWP) to the Indian Ocean during this time, resulting in this early Miocene CCD deepening. Episodes of higher nutrient concentration favoring carbonate productivity (small-size coccolith production) in the western tropical Pacific over the Miocene correspond to cooling and a shallower thermocline affected by the Western Pacific Warm Pool (WPWP) dynamics (Imai et al., 2015; Yuwono and Sato, 2020). During the early Miocene, when the Indonesian Seaway was still open, the Indo-Pacific water exchange and westward WPWP flow to the Indian Ocean created a shallow thermocline across the western tropical region (Von Der Heydt and Dijkstra, 2011), which could possibly contribute to carbonate accumulation. The role of the thermocline in calcareous biogenic productivity across the western tropical and western North Pacific during the Miocene and Pliocene (Imai et al., 2015; Pratiwi and Sato, 2016) suggests that it may have played a major role in driving CCD variations in the western tropical Pacific since the Miocene. For instance, the cooling and shoaling of the thermocline and enhanced water mixing across the western tropical Pacific since 4 Ma, with intensified episodes between ∼3.1 Ma and ∼0.7 Ma (Dang et al., 2024), agree with our modelled CCD deepening from ∼3.8 km at 3 Ma to ∼4.5 km at 0.5 Ma in this region. The western tropical Pacific thermocline and its vertical changes, affected by the East Antarctic glaciation and the progressively reduced Indo-Pacific water exchange through the Indonesian Seaway constriction during the Miocene (Gallagher et al., 2009; Gasperi and Kennett, 1993; Nathan and Leckie, 2009), and its effect on the CCD variability of this region require further investigation.

Although Imai et al. (2015) described a similar effect of the thermocline changes on carbonate production in the western North Pacific over the Neogene, the inferred timing by the authors (e.g., shoaling of the thermocline since 5 Ma) does not exhibit a clear pattern correlative with our western North Pacific CCD fluctuations. In the middle Miocene, a CCD shoaling in the western North Pacific, peaking at ∼3.5 km between 15 Ma and 14.5 Ma (Figure 3B), could be associated with the Miocene Climate Optimum (MCO: 17–14 Ma; Westerhold et al., 2020). Holbourn et al. (2014) used benthic foraminiferal isotope signatures from the eastern equatorial Pacific during the MCO combined with astronomically tuned δ¹⁸O and δ¹³C proxies from the southeast and northwest Pacific covering the same interval from an earlier study (Holbourn et al., 2007) to suggest that the Southern Hemisphere warming led to an ∼5 °C increase in Pacific deep ocean temperature between 15.7 and 14.7 Ma, which in turn resulted in enhanced deep-sea carbonate dissolution and a CCD rise. According to the authors, carbonate preservation recovered during the next cooling phase at ∼13.8 Ma, associating with Antarctic ice sheet expansion, which correlates with a reduction in atmospheric CO₂ (Hönisch et al., 2023). This pattern and subsequent recovery are reflected in a CCD shoaling within MCO, peaking at ∼3.5 km between∼15 and ∼14.5 Ma, followed by a deepening to ∼4 km at 13.5 Ma in our model from the western North region. A recent study by Holbourn et al. (2024) based on a multi-proxy record from Site U1490 in the western equatorial Pacific shows an increased carbonate accumulation over the MCO followed by a decrease during the Middle Miocene Climate Transition (MMCT). These excursions agree with our CCD model for the western tropical Pacific (Figure 4B), which shows a CCD deepening from ∼3.6 km at ∼16.5 Ma to ∼4.1 km at its end, followed by a short-lived CCD shallowing to 3.5 km at the start of the MMCT. Holbourn et al. (2024) attribute the carbonate accumulation high during the MCO to increased primary productivity, possibly related to an increase in weathering (Bolton et al., 2016; Si and Rosenthal, 2019), which compensated for increased carbonate dissolution related to the increased formation and northward expansion of corrosive deep waters in the western tropical Pacific. During the MMCT, carbonate preservation improved due to reduced CO₂ uptake at deep-water formation sites in the Southern Ocean, while carbonate accumulation decreased (Holbourn et al., 2024) suggesting that primary productivity also decreased at this time.

Since the late Miocene, the CCD fluctuated between ∼3.4 km and ∼3.9 km in the western North Pacific (Figure 3B), which could be linked to the impact of the Panama Gateway constriction on the deep-water chemistry of the Pacific. At present, the Pacific deep water is becoming enriched in dissolved silica, while displaying relatively low carbonate ion concentrations as deep water flows and ages toward the North Pacific (Zhang et al., 2022). This northward silica gradient across the Pacific mirrors the late Miocene pattern, following the formation of AABW and its northward flow to the North Pacific (Woodruff and Savin, 1989). This silica gradient is triggered by reduced Atlantic-Pacific water exchange through the restriction of the Panama Gateway and enhanced nutrient concentration and productivity across the eastern equatorial Pacific (Schneider and Schmittner, 2006). Productivity increases result in opal sedimentation due to faster rain of opal and its higher preservation compared to carbonate (Lyle and Baldauf, 2015). Recycling and dissolution of biogenic opal before complete burial (Ragueneau et al., 2000) could result in enhanced dissolved silica (DSi) toward the North Pacific. Additionally, Panama Gateway constriction enhanced the North Atlantic Deep Water (NADW) production (Butzin et al., 2011; Schneider and Schmittner, 2006) and accompanied by Antarctic cooling, led to an increased AABW flow to the Pacific enriched in dissolved silica (Keller and Barron, 1983), triggering the DSi pool to the North Pacific (Cortese et al., 2004). All these occurrences have resulted in enhanced silica burial in the western North Pacific over the late Neogene (Barron, 1998; Rea et al., 1995), which is reflected in the overall shoaling trend of our CCD model since the late Miocene. We interpret this CCD rise in relation to silica flux because diatom productivity surpasses coccolith accumulation under relatively high nutrient supply conditions (Litchman, 2007), and higher silica productivity intervals coincide with carbonate dissolution (Lyle and Baldauf, 2015), which leads to a shoaling trend of the CCD. This is supported by the stratigraphy of ODP sites 881, 882, 883 and 884 from the western North region (Figure 1), which is dominated by post-late Miocene diatom ooze. In addition, our CCD reconstruction in the western North Pacific shows shoaling peaks at 6.5 Ma and 3 Ma, which correlate with two notable peaks of silica accumulation (at ∼6 Ma and ∼3 Ma) in this region (Rea et al., 1995).

Our eastern equatorial CCD model shows two shallowing pulses in relation to two major events of the late Miocene, encompassing the carbonate crash and biogenic bloom, in this region (Figure 4A). The first shoaling, with an onset at ∼11.5 Ma, corresponds to the carbonate crash event recognized by Lyle et al. (1995) in different parts of the eastern equatorial Pacific between 11 and 8 Ma. Lyle and Baldauf (2015) determined that this event occurred between 13.2 and 8 Ma in the Pacific eastern equatorial region. Our CCD reconstructions suggest that the peak of the carbonate crash occurred between 10.5 and 9 Ma over a regional scale across the eastern equatorial Pacific, coinciding with the lowest carbonate preservation of ∼9.7 Ma from Lyle et al. (2019). This event is reflected in the CCD shoaling of ∼800 m from ∼4.8 km at 12 Ma to ∼4 km at 10 Ma in our model, ∼300 m shallower than the value computed by Pälike et al. (2012) for this period (Figure 4A). The carbonate crash may be explained as a consequence of the Panama Gateway constriction by enhanced carbonate dissolution due to intensified AABW flow and corrosive water circulation to the Pacific (Lyle et al., 1995; Roth et al., 2000). Furthermore, in the western tropical Pacific, the CCD shoals by ∼900 m, from ∼4.2 km at ∼11 Ma to ∼3.2 km at ∼8 Ma (Figure 4B), indicating the occurrence of the carbonate crash event in this region. This marked shift further suggests that the equatorial region of the Pacific has experienced a substantial response to the carbonate crash event during the late Miocene. Based on our CCD data, we propose the initiation of carbonate crash in the eastern equatorial Pacific at ∼11.5 Ma and in the western tropical Pacific at ∼10.5 Ma. Additionally, the CCD shallowing corresponding to the carbonate crash peaks at ∼8 Ma in the western tropical region, compared to that of the eastern equatorial of ∼10–9.5 Ma. These patterns may offer the east-west lag in the onset and intensification of the carbonate crash. Nathan and Leckie (2009) described a ∼1 Ma delay in the initiation of the carbonate crash event from the Caribbean to the eastern equatorial Pacific, which we argue may have further extended into the western tropical region of the Pacific. However, considering the 0.5 Ma resolution of our analysis, this interpretation should be regarded with caution. We also note a signal of this event in our model for the eastern South Pacific expressed by a rise of the CCD from ∼3.5 to ∼3.2 km between 12 and 10.5 Ma (Figure 6A). The carbonate crash event was also inferred in the eastern South Pacific by Preiss-Daimler et al. (2021) based on a carbonate accumulation decline at ∼11 Ma. The occurrence of the carbonate crash event in the eastern South Pacific suggests that the oceanographic effect of the Panama Gateway constriction extended beyond the Pacific equatorial region. However, our findings indicate that the carbonate crash may not have occurred as a basin-wide event across the Pacific Ocean. The spatially variable expression of the carbonate crash within the Pacific broadly corresponds to the carbonate crash signals observed in the low-to mid-latitude Atlantic and Indian Oceans (Lübbers et al., 2019; Preiss-Daimler et al., 2021). Notably, our CCD model reveals no evidence of this occurrence in the western North Pacific, where the CCD remained relatively stable between ∼3.9 and ∼3.8 km from ∼12 to ∼8.5 Ma (Figure 3B). Moreover, examining core sediments from deep-sea drill sites in this region, like those from the Shatsky Rise (Figure 1), which extend into the middle Miocene, demonstrate well-preserved carbonate sequences predominantly composed of nannofossil ooze, with CaCO₃ concentrations averaging ∼60–70% during this interval.

The subsequent CCD shallowing in the eastern equatorial region of the latest Miocene, peaking between ∼7 and ∼6.5 Ma (Figure 4A), corresponds to the late Miocene biogenic bloom event of the eastern equatorial Pacific, which may be another response to the Panama Gateway constriction and the resulting differentiation in Atlantic-Pacific deep water (Farrell et al., 1995; Lyle and Baldauf, 2015; Lyle et al., 2019; Pillot et al., 2023). The occurrence of the biogenic bloom from our model is coeval with the late Miocene carbon isotope shift between 7.7 and 6.4 Ma (Drury et al., 2017). While the underlying mechanisms of the late Miocene biogenic bloom remain debated, Pillot et al. (2023) argue that the heterogeneous distribution of biogenic bloom (with peak intensity at ∼7 Ma) across the global ocean may be driven by the strengthening of thermohaline circulation or intensified wind-driven upwelling on a global scale, both influenced by climate variability and gateway dynamics. Here, our CCD record correlates with the highest biogenic opal accumulation rates recorded between 7.1 and 6.4 Ma from Lyle and Baldauf (2015), and between 7 and 6.5 Ma from Lyle et al. (2019) in the eastern equatorial region. Lyle et al. (2019) suggest that despite high carbonate burial since 14 Ma (apart from the carbonate crash interval), enhanced diatom productivity during the biogenic bloom was accompanied by reduced carbonate accumulation across the equatorial high-productivity region of the Pacific. Supporting this, Lyle et al.’s (2019) data indicate that an average of ∼20% increase in biogenic opal wt% corresponded with a ∼20% decrease in CaCO₃ wt% between ∼7 and ∼6.5 Ma at drill site U1338, which is part of a cluster of eight drill sites, identified within the biogenic bloom manifestation zone in the eastern equatorial Pacific (Pillot et al., 2023). Within this period, opal wt% peaked at ∼39%, while carbonate content declined to ∼44%, representing a ∼35% reduction from ∼7.5 Ma. These observations may suggest that enhanced opal accumulation is captured in our CCD model by a peak of shoaling (∼7–6.5 Ma) across the eastern equatorial region. However, in the western tropical region, CCD shoaling reached its peak between ∼5.5 and ∼5 Ma (∼3.3–3.2 km), when the eastern equatorial CCD deepened to ∼4.5 km by ∼5 Ma (Figures 4A,B). The observed offset in CCD trends between the western and eastern equatorial Pacific until ∼4 Ma may reflect perturbations between biogenic opal production and carbonate accumulation, modulated by thermocline variability across the equator. The thermocline shifts, influenced by the Pacific warm pool dynamics, contributed to longitudinal asymmetry in species distribution across the equatorial Pacific during the late Miocene (Nathan and Leckie, 2009). Our results also represent a subsequent relative convergence of CCD patterns across the western tropical and eastern equatorial regions between ∼4 Ma and ∼1 Ma, which coincides with a diminished west-east thermocline gradient (Wara et al., 2005) and the disappearance of a nannofossil content gradient across the equatorial Pacific during this interval (Bolton et al., 2010). Nevertheless, a more comprehensive understanding of the interplay between CCD variability, Pacific warm pool fluctuations, and the differential contributions of opal and carbonate productivity across the equatorial Pacific necessitates further examination of high-resolution proxy data.

We present new models of the CCD across the entire Pacific Ocean, split into six regions (western North Pacific, eastern North Pacific, western tropical Pacific, eastern equatorial Pacific, western South Pacific, and eastern South Pacific). Our results demonstrate distinct regional variability in CCD fluctuations across the Pacific, which differ from the reconstructed average CCD that characterizes the overall Pacific basin. We find the deepest CCD in the eastern equatorial area, ranging between ∼3.7 and ∼4.7 km prior to the middle Miocene. In contrast, the western North Pacific region features a relatively shallow CCD (∼2.3–4 km) over the majority of the Neogene, rarely exceeding 4 km. Coupled with the diverse CCD changes between the eastern equatorial and western tropical regions of the Pacific, our CCD models highlight substantial latitude-longitude heterogeneity in carbonate flux across the Pacific Ocean over the late Cenozoic. We interpret the CCD changes across the Pacific regions to reflect the effect of climate variability, the Antarctic glaciation, and the configurations of the Indonesian and the Panama gateways on regional carbonate productivity across the Pacific over the Neogene. Our model of the western tropical Pacific CCD, the first of its kind, demonstrates a distinct drop at the earliest Miocene that may be related to marine West Antarctic ice sheet growth at 26 Ma (Klages et al., 2024), suggesting a delayed deep ocean cooling after ∼2 Ma. We propose the peaks of CCD shoaling ∼10–9.5 Ma and ∼7–6.5 Ma corresponding to the carbonate crash and biogenic bloom events, respectively, in the eastern equatorial Pacific. These occurrences are also observed in the western tropical region from our CCD models, which may imply these events are characterized over the equatorial region of the Pacific.

Our study provides constraints on CCD perturbations on regional scales. The particularly deeper CCD of the eastern equatorial Pacific, compared to the western tropical and western North regions during the Miocene, offers a caution against the common practice of using the eastern equatorial CCD to represent the entire Pacific (e.g., Bogumil et al., 2024; Boudreau and Luo, 2017) in long-term carbon cycle modeling. However, there is a vital need for future drilling targeting oceanic crust older than 60 Ma in the western Pacific to gain deeper insights into CCD behavior during the warm intervals of the Cretaceous and Paleogene. Additionally, future research could focus on better constraining the CCD and carbonate accumulation in other regions of the global ocean, noting that their evolution is poorly-defined in the Indian and Southern oceans.