Depositional patterns constrained by slope topography changes on seamounts

Slope topography is known to control the spatial distribution of deposits on intraplate seamounts; however, relatively little is known about how slope topography changes constrain those depositional patterns. In this study, we analyse data on four lithotypes found on seamount slopes, including colloidal chemical deposits comprising mainly cobalt-rich crusts, and examine the relationships between the spatial distribution of these lithotypes and current slope topography. We use these relationships to discuss depositional patterns constrained by slope topography changes. Some depositional units in drill core samples are interpreted to have resulted from past topographic changes that created the current slope topography. Two or more types of deposits that accumulated at the same location implies that the slope topography changed over time and that the depositional patterns on seamount slopes are constrained by changes in slope topography.

. The Magellan Seamounts and locations of geological survey sites on them. (a) The map images are created using the SRTM30_PLUS 13 . Surveys acquired 205 shallow boreholes, including 82 from seamount MA, 51 from seamount MC, 37 from seamount ME and 35 from seamount MK, together with 20 TV grabs and 161 dredging samples on slopes between 1200 and 3500 m deep. Shallow drill sites are indicated by white dots shown in (a). Bathymetric data on the four seamounts covering an area of approximately 40,000 km 2 measured using the EM122 Multibeam Survey System (Kongsberg, Inc.) by several survey cruises of the COMRA. Topography image of the seamount ME created using the bathymetric data is shown in (b). The TV video surveying line covering 18 km and two sizeable landslides on the flank of seamount ME are shown in (b), will be involved in the following.

Scientific Reports
| (2020) 10:20534 | https://doi.org/10.1038/s41598-020-77573-2 www.nature.com/scientificreports/ little is known about how changes in slope topography constrain depositional patterns over the geological history of the seamount. Over the past 10 years, the China Ocean Mineral Resources Association (COMRA) has conducted several survey cruises to study the Magellan Seamounts ( Fig. 1) and has accumulated a large amount of data that are the focus of this study ( Fig. 1; "Methods" section).
Four lithotypes. Four distinct lithotypes exposed on the seafloor have been identified in this study. (1) Exposed rock such as volcanic breccias, basalt and consolidated sediments 34 are exposed on seafloor and are not covered by any deposits. (2) Loose sediments are soft and unconsolidated and mainly comprise clay, silt and foraminiferal sands. These are interpreted as pelagic sediments, and are easily mobilised downslope by gravity processes 32,35,36 . (3) Cobalt-rich crusts comprise chemical colloidal deposits enriched in cobalt, phosphorite or calcium 37,38 . These occur as crusts attached to substrate rock surfaces that are relatively stable. (4) Transitional sediments accumulate on the transition zone separating slopes of different gradients. They are an unconsolidated mixture of rock debris, breccia, gravel, sand, clay, cobalt-rich crusts and ferromanganese nodules, as previously described by Yamazaki and Sharma 15 . Compared with loose sediments, transitional zone deposits experience a lower degree of mobilisation on seamount slopes. Similar observations have been presented by Yeo et al. 25 . All the lithotypes in this study are shown in Fig. 2. Except for exposed rock, each lithotype is a distinctive depositional facies.
Although the loose sediments on the flat tops of seamounts are similar to pelagic sediments everywhere, transitional zone sediments and cobalt-rich crusts on seamount slopes have distinct depositional patterns when compared with deep-sea basin and continental margin sediments. Additionally, the spatial distribution of the loose sediments, the transitional zone sediments and cobalt-rich crusts on seamount slopes are mainly constrained by seamount topography (i.e. slope gradient is a key factor).

Results and discussion
Slope gradients of lithotypes. The datasets involved in this study were obtained by geological sampling (e.g. shallow drilling and TV grabs) from 225 geological survey sites (Fig. 1a). Each dataset of the four lithotypes-loose sediments, transitional zone sediments, cobalt-rich crusts and exposed rock, includes information on the depth and slope gradient of the survey sites (Fig. 3). As shown in Fig. 3, loose sediments are generally confined to gentle slopes, whereas exposed rock and cobalt-rich crusts are preferentially distributed on steeper slopes and transitional zone sediments tend to appear on slopes with gradients between those characteristic of loose sediments and cobalt-rich crusts.
Similar observations can be found in the literature, and previous studies have identified slope topography's role in controlling the spatial distribution of depositional patterns on seamount slopes. For instance, cobalt-rich crusts commonly form ring-shaped deposits on upper seamount slopes and along the periphery of the summit 12 . This contrasts with the flat tops and sharp escarpments of underwater seamounts, where the slope is less favourable for accumulating such deposits 26 . Typical seamount slopes can be divided into three zones: one dominated by nodules which are referred to one of the components of the transitional zone sediments here, with a gradient of up to 4°, one dominated by sediments with gradients of up to a 3°, and one with cobalt-rich crusts with gradients of up to 15°1 5 . Shallow profilers and TV video have been used to delineate the upper boundaries for cobalt-rich crust deposition 27,39 . This is in addition to studies using shallow drilling and TV video to identify how micro-topography controls cobalt-rich crust enrichment 28 . Slopes greater than 20° are not suitable for growing cobalt-rich crusts 29 . However, slopes with cobalt-rich crusts all have gradients greater than 4° and slopes with loose sediment are all less than 4°3 0 .
Deviations exist between different observations that could be caused by measurement error and cumulative influence of bottom flow, as elucidated by Stow et al. 23 ; however, these observations consistently imply that current slope gradient controls the spatial distribution of lithotypes on slopes. Additionally, exposed rock carries other geological implications that will be explored in the next section.
Intense slope topography changes and the depositional record. Huge slope collapses or landslides on the flank of seamount ME described by Smoot and King 40 are typical topographic changes observed on seamounts around the world, and these events can be identified by several features 41 . The survey data shown in Table 1 provide a basis for evaluating landslide influence on slope depositions. Two drag samples (D04 and D05) and a shallow drilling sample (ME9) were taken from the body of landslide 1, as shown in Fig. 1b. Sample D04 comprises volcanic breccia with a manganese film, sample D05 consists of volcanic breccia and ME9 contains basalt without chemical encrustation. Two other drag samples (D06 and D07) and a shallow drilling sample (ME38) were taken from landslide 2, as shown in Fig. 1b. Sample D06 comprises volcanic breccia, sample D07 consists of limestone and ME38 contains volcanic breccia without chemical encrustation. The data reveals that besides residual slope depositions, the lithotypes underlying the landslides are primarily exposed rocks.
Intense topographical changes on the slope, such as landslides, slope collapse and volcanic activity (which ceased in the Late Cretaceous in this case) on intraplate seamounts 10,42 , play a role in clearing deposits from seamount slopes and burying old deposits on slopes 25 . After past deposits have been removed, the seamount flank would consist exposed rock and would eventually experience build-up of new deposits.
Obviously, slope topography changes induced by either sudden collapse or gentle adjustments in slope gradient can influence deposition on the slope. For instance, landslides remove material from the slope in their proximal part and cover or disturb deposits in their distal part. When slopes undergo gentle adjustments in gradient, they become steeper, promoting growth of cobalt-rich crusts, or become gentler, inhibiting such growth. www.nature.com/scientificreports/ Therefore, this observation can be used as a basis to explore interactions between changes in seamount topography and deposition on seamount slopes in the geologic record.
A stratigraphic record of slope topography changes. A core was obtained from the shallow borehole site labelled ME01 in Fig. 2a. Its section image is shown on the right side of Fig. 4a. In this sample, there is an approximately 5 cm thick consolidated intercalation of breccia, debris, ferromanganese nodules and gravel Mel'nikov et al. 43 referred to this intercalation as buried ferromanganese nodules. In our work, we refer to this intercalation as the breccia layer between cobalt-rich crust and substrate rock. Twenty-four of the 205 drilling samples from the Magellan Seamounts, including six from seamount ME shown in Fig. 1b, have a breccia layer similar to ME01. Most importantly, those breccia layers have the same composition as transitional zone sediments and are 3-30 cm thick. Thus, we interpret the breccia layer as being buried transitional zone sediments. The stratigraphic relationship of cobalt-rich crust, which prefers larger gradient slopes, accumulated on transitional zone sediments, which prefer transitional gradient slopes (Fig. 3), observed in one drill core, suggests that the seamount slope topography has experienced adjustment over its geological history. The lack of loose sediments in the drill core suggests that if they were initially present, they have been lost due to mobilisation during the slope adjustment that resulted in the stratigraphic formation of cobalt-rich crust and transitional zone sediments.
Stratigraphic chronology and records of slope changes in cobalt-rich crusts. There are layered structures in the cobalt-rich crust sections, e.g. layers III, II, I2 and I1 in Fig. 4c-e. Each layer has distinct     Fig. 4b,c were obtained underwent a gentle slope adjustment, going from a gentle slope to larger gradient slope. By contrast, the slopes where the drill cores shown in Fig. 5d,e were Table 2. Stratigraphic chronology of cobalt-rich crusts on the Magellan Seamounts. *Samples obtained through drilling shallow boreholes and dredging, whole rock analysis by absorption spectrometry. **Estimated by the formula of Puteanus and Halbach 49 . ***According to Melnikov and Pletnev 50 .
Dating by growth rate and thickness of cobalt-rich crust** Dating by cobalt-rich layers***  www.nature.com/scientificreports/ obtained underwent sudden slope collapse, becoming steep and providing clear earlier deposits for the growth of subsequent cobalt-rich crusts. The slope where the drill core shown in Fig. 4b is located maintained a transitional zone slope topography for long time (approximately 5 Myr) and underwent deposition of transitional zone sediments that have since been consolidated. The slope where the drill cores shown in Fig. 4c are located is similar to that of ME01 and underwent a sequential slope gradient adjustment allowing deposition of transitional zone sediments and cobalt-rich crust. The slope where ME10 is located (Fig. 4e) underwent sudden slope collapse, forming a steep slope for long term (approximately 53 Myr), for Layers III, II, I2 and I1 to deposit on. The slope where ME43 sits (Fig. 4f) was deformed to an 18.9° gradient slope. This slope is too steep for loose sediments and transitional zone sediments to deposit on and was exposed to the seafloor too recently to allow ferromanganese film deposit.

Seamount Sample numbers* Co (%) Growth rate (mm/Myr) Crust layer Age (Myr) Stratigraphic chronology
Seamount evolution and slope topography changes. The seamounts underwent several volcanic eruptions before reaching the extinct stage, causing different rock compositions to stack layer by layer, which generally formed the current seamount topography (Fig. 5). It suggested that the first stage layers consist mainly of basalts and that the second stage layers are made up of basalt and 20% volcaniclastics 51 . Volcaniclastics are the dominant rock types 52 in the third stage, accounting for more than 60% and are deposited on flanks in the last stage 10 . When seamounts entered the extinct stage, subsidence due to the load of the seamounts themselves became the dominant vertical force 53 . For instance, the island of Hawaii has a subsidence rate of 2.6 mm/year 54,55 . When the seamounts subsided below sea level, biogenic limestone began being deposited during the Cretaceous (Aptian-Cenomanian), which continued for a long time. This limestone is 470-800 m thick and is mainly distributed between 800 and 2500 m deep 56 (Fig. 5). Over time, seamount surfaces became encrusted with chemical colloid depositions, mainly cobalt-rich crusts. Meanwhile, pelagic sediments (referred to as loose sediments) were deposited on the slopes and mainly on the flat tops of summits (Fig. 5a). Intrusive activities on seamounts are demonstrably major factors in their evolution, leading to slope collapse on seamount flanks 10 and topography adjustment in slope gradients caused by plastic deformation [57][58][59] or seamount subsidence.
We suggest that limestone on seamount slopes, especially on the summit boundaries (e.g. sites where the six cores containing breccia layers were collected on seamount ME shown in Fig. 1b) underwent plastic flow as described in previous studies [57][58][59] . Limestone had been moving down the slope under the pull of gravity at a very slow speed, e.g. 0.1 mm/year (the speed will be discussed in next paragraph). When plastic flow occurred, the limestone and overlying deposits (i.e. transitional zone sediments and cobalt-rich crusts) moved down the slope together, and some micro-deformation structures would be formed in the overlying rocks 60 . The microdeformation structures in a cobalt-rich crust shown in Fig. 5d could be evidence of plastic flow. When the plastic flow moved to a different part of the slope, the gradient was altered accordingly.
As suggested by Fig. 5c, before 5 Myr, the limestone rock with overlying transitional zone sediments underwent plastic flow downslope from site Pic3 to site ME01 over a long period (up to 5 Myr), and the slope changed from a transitional gradient (a ≈ 4.8° ± 1.2°) to a steeper gradient (a > 4.8 ± 1.2°) accordingly, with cobalt-rich crust deposits overlaying the transitional zone sediments. The distance between sites Pic3 and ME01 is approximately 500 m; therefore, if the plastic flow took 5 Myr, then the plastic flow velocity can be estimated to be 0.1 mm/year.
Another factor involved in slope gradient changes could be unequal subsidence between different rock composition layers (right panel of Fig. 5a).
When collapse occurs on the flanks, previous deposits would be either cleared or covered, and bare rocks would be exposed to the seafloor, allowing subsequent deposition to develop. On steep slopes, old landslides deposited thicker cobalt-rich crusts (left panel in Fig. 5b) and later landslides deposited thin cobalt-rich crusts or exposed bare rocks to the seafloor without overlying deposits (right panel in Fig. 5b).

Model for depositional patterns constrained by topography changes.
(1) Seamounts need to have gentle slopes to promote loose sediment deposition (referred to as gentle slope for loose sediments). Two types of deposit (i.e. loose biogenic particles and ferromanganese chemical colloids) are well preserved on gentle slopes or on the flat tops of seamounts with gradients less than the threshold value (a < a 0 ). Loose biogenic particles are generally larger and are more matter flux, whereas ferromanganese chemical colloids are smaller and less matter flux. Therefore, these two types of sediment deposited on the gentle slopes together, with the former dominating the latter, and the depositional pattern presents as loose sediments (Fig. 6a). (2) When slope gradients achieve a threshold value (a ≈ a 0 ), transition zone slope development, in which transitional zone sediments are deposited (referred to as transition zone slope for transitional zone sediments) can occur. Loose particles would be subject to higher rates of mobilisation and could be lost. By contrast, breccias, ferromanganese nodules and gravels that exhibit less tendency for mobilisation would remain on the slope. If the slope maintained this topography for a long time, the breccias and nodules would consolidate through some type of cementation (e.g. carbonate cementation), at which point transitional zone sediments would be preserved in the stratigraphic column. However, if slopes increase over time, transitional zone sediments do not have enough time to consolidate and will be removed. This pattern is shown in Fig. 6b. (3) Slopes with larger gradients (a > a 0 ) promote deposition of cobalt-rich crust (referred to as larger gradient slope for cobalt-rich crust). Larger gradient slopes encourage the loss of loose sediments as their sliding friction is not enough to resist the pull of gravity. This exposes stable substrates, promoting chemical colloid deposition, mainly cobalt-rich crusts. This pattern is shown in Fig. 6c.
Scientific Reports | (2020) 10:20534 | https://doi.org/10.1038/s41598-020-77573-2 www.nature.com/scientificreports/ (4) Intense slope topography collapses such as those resulting from landslides, slope collapse and volcanic activity could reset seamount slopes (referred to as slope resetting for new depositions). This includes the removal or burial of prior deposits followed by deposition of new sediments. This pattern is shown in Fig. 6d. (5) The above scenarios are various static slope deposition patterns that can occur in different orders and combinations on a single slope over its geological history. Thus, the depositional patterns constrained by changes in topography would be explained by a combination of these scenarios. This model could be used to explain the depositional sequences observed in sections of samples taken from seamount slopes.
Several applications of the models.
(1) Slope topography has changed over time at the location of ME01. The section from sample ME01 shown in Fig. 4a has three stratigraphic layers, namely, an overlying cobalt-rich crust, a breccia layer and an under- www.nature.com/scientificreports/ lying substrate rock layer. The cobalt-rich crust in layer III is 25 mm thick, suggesting that its formation began in the Pliocene. We thus speculate that the following process occurred. Approximately 5.3 Ma, the slope where ME01 is in Fig. 2a was a gentle slope or flat top seamount covered with loose sediments (gentle slope for loose sediments). The slope then changed from a gentle slope to a transition zone slope, which was accompanied by the loss of most of the finer grained components, lithification of larger particles and breccia rocks through chemical colloid deposition and burial of transitional sediments as the breccia layer (transition zone slope for transitional zone sediments). Lastly, the slope gradient continued to increase, resulting in deposition of a cobalt-rich crust (larger gradient slope for cobalt-rich crust). These processes are shown in Fig. 6f. (2) Explanation of some sections in core samples Additional processes such as larger gradient slope for cobaltrich crust => gentle slope for loose sediments shown in Fig. 6e could explain hidden cobalt-rich crusts under loose sediment, and larger gradient slope for cobalt-rich crust => gentle slope for loose sediments => larger gradient slope for cobalt-rich crust might provide an alternative explanation for the interrupted growth of crusts found by Klemm et al. 61 and Meng et al. 62 .
When slope topography changes occurred, the uppermost deposits might also change and be preserved within seamount slope stratigraphic records. Therefore, stratigraphic information obtained through drill cores might provide evidence of the geological history of slope changes. The models shown in Fig. 6 could thus be used to interpret sections of drill cores from slopes on all intraplate seamounts, including guyots and spire seamounts.

Methods
Positioning methods. A combination of on-board GPS and underwater acoustic positioning techniques allowed for a spatial resolution of relative position of up to 5 m.
Bathymetry surveying and data processing. Bathymetric data were collected using an EM122 (Kongsberg, Inc.), a multibeam survey system that generates data that enables the production of wide-swath contour maps of the seafloor. Bathymetric data for all survey lines were processed manually on-board using Seafloor Information System software version 3.6. Post-processing consisted of editing the cross-track and navigation data (including the deletion of bad data, correction of position, etc.), leading to the creation of grid data. A 100 × 100 m grid size was selected because the raw data's horizontal resolution was approximately 45-75 m, depending on the interval water depth of the studied region (1500 to 2500 m). The accuracy of the raw depth data is 3.0-5.0 m based on the following equation: depth accuracy = depth × 0.2%. Seafloor TV video. The seafloor TV video technology constructed by COMRA was used to obtain images of the seafloor along a 15 km survey line along the southeast corner of seamount ME. The video camera was situated around 3-5 m from the seafloor and the video focus window on the seafloor is approximately 3 m × 5 m.

Estimating slope gradients.
Cobalt growth rate. Cobalt concentration analysis was conducted using absorption spectrometry. Thus, the age and growth rate of each cobalt-rich crust layer can be recalculated using the following formula: growth rate (mm/Myr) = 1.28/[Co (%) − 0.24] 49 . Related data are shown in Table 2.
Drilling shallow borehole samples. The shallow boreholes were drilled using submersible rigs constructed by COMRA. Samples were obtained from 205 stations the distribution of sampling sites is shown in Fig. 1. Core sizes are 300-1000 mm long and 50 mm in diameter. Core samples were classified, and the thickness of their crusts was measured in the laboratory. Statistical parameters of the data are shown in Fig. 3. Section images of cores. Sample cores were divided into two parts with a chainsaw and then polished and cleaned with fine sandpaper. Section images were captured using a digital camera. The image of microdeformation structure in cobalt-rich crust is taken with a Zeiss microscope (Axioskop 40).

Data availability
The authors declare that the main data supporting the findings of this study are contained within the paper and are available in the Mendeley data (http://doi.org/10.17632 /g8359 3d5jc .3). All other relevant data are available from authors upon reasonable request.