The studied samples were obtained from the Global Change and Air-Sea Interaction Special-West Pacific Block PAC-CJ15 Submarine Sediment and Benthic Organisms Investigation Project. The samples were collected using a box corer from April to June 2017 (Zhang et al., 2021; Zhou et al., 2021). We selected 80 upper sediment samples (0-10 cm) for analysis, and the samples contained sediments since the Holocene referring to the sedimentation rate near the study area (Cai, 2019). The distribution of sampling stations is shown in Figure 1B .

All samples were processed in the diatom analysis laboratory of the Third Institute of Oceanography, Ministry of Natural Resources, Xiamen, China. We prepared samples following a modified version of the method of Hakansson (1984). Briefly, samples weighing ca. 5 g were dried in an oven at 60°C and weighed before processing. Samples were treated with 10% HCL and 30% H₂O₂ to remove carbonates and organic material, respectively, and washed in distilled water to remove these chemicals from the solution. Samples were then soaked in distilled water for 24 h and scattered using an ultrasonic dispersion instrument (120 Hz) for 2 min. Diatoms (>15 μm) were the main components, which indicates their relationship to the environment would be reliable. Small diatoms (<15 μm) have been shown to little effect on the results of this type of study, so were not considered (Chen et al., 2019). Therefore, samples were filtered through a 15-µm sieve to remove microdiatoms and other material finer than 15 µm, after which the suspension containing diatoms was concentrated to 2 ml. When completely homogeneous, the sub-sample of the suspension was transferred to a cover slip and air-dried. Finally, three permanent slides were made for each sample using Canadian balsam as a fixative (Chen et al., 2020b). The prepared slides were observed and identified under an optical microscope (Olympus BX51, objective lens 40×, ocular lens 20×), and at least 300 diatom valves were counted for each sample. For incomplete individuals, valves were counted if they were more than half complete. Diatom classification and identification mainly followed illustrations from published studies (Jin et al., 1982; Round et al., 1990; Jin et al., 1991; Cheng et al., 1996; Guo and Qian, 2003; Qi and Li, 2004; Smol and Stoermer, 2010; Cheng and Gao, 2012). Because Ethmodiscus rex specimens in the samples were fragmentary, we performed only area-based qualitative statistics.

The total abundance of diatoms is the total number of valves per gram of air-dried sediment (valves/g). Species diversity was calculated using the Shannon-Weaver index (Shannon and Weaver, 1949). The diatom assemblages were statistically analyzed using the R software environment and correlated with various environmental factors. Prior to statistical analysis, data were first standardized for all species. In the analysis, species with >2% relative abundance in at least one sample were included (Imbrie and Kipp, 1971), and rare species (those occurring at fewer than five stations) and data for station E13 (where no diatoms were found) were excluded. Diatom assemblages were differentiated using cluster analysis. Since detrended correspondence analysis showed a maximum gradient of 1.68 < 2, redundancy analysis (RDA) was used to determine whether differences in the assemblage vectors of each sample were statistically significant.

The environmental data obtained in this study included sea-surface temperature (SST), salinity (SSS), dissolved oxygen (DO), nutrients (silicate, phosphate and nitrate), chlorophyll a, net primary production (NPP), particulate organic carbon (POC) and water depth (WD). Data such as SST and SSS were obtained in situ using the SBE 911Plus Conductivity-Temperature-Depth system; nutrient and oxygen levels were derived from the CARS2009 dataset for 1930 to 2009 with 1/2°-resolution (Qu et al., 2022); chlorophyll a and POC concentrations for 1997 to 2009 at 9-km resolution were downloaded from https://oceancolor.gsfc.nasa.gov/l3/; NPP data for 1997 to 2009 with 1/6°-resolution were downloaded from http://www.science.oregonstate.edu/ocean.productivity/custom.php; and WD data in those figures from the GEBCO-2020 Grid at 15 arc-second intervals were downloaded from https://data.noaa.gov/onestop.

According to the data obtained from in situ measurements and websites ( Figure 2 ), SST at each station varied relatively little, with a range of 28.90 to 29.27°C (average 29.12°C). The SSS variation was also small, from 33.94 to 34.07 (average 33.99) whereas the DO content varied from 4.52 to 4.56 ml/l (average 4.54 ml/l). Nitrate and phosphate contents varied little, ranging from 0.03 to 0.06 µmol/l and 0.04 to 0.11 µmol/l, respectively, whereas silicate content varied widely (1.11-2.97 µmol/l) and showed a trend of gradual increase from the northeast to the southwest. This increase occurs because the study area is located in the main pathway for the invasion of the silica-rich AAIW into the Northwest Pacific Ocean (Reid, 1973; Zhang, 2008; Zhang et al., 2016). In addition, NPP, POC and chlorophyll a content all showed a gradual increase from north to south, indicating higher phytoplankton biomass and primary production in the southern part of the study area.

A total of 53 species or varieties of diatoms belonging to 22 genera were identified at 80 stations. The relative percentages of each taxa were listed in the table of supplementary material. Diatom abundance varied from 0 to 88,373 valves/g, with an average of 14,319 valves/g. Some individual stations (E13, C03, B04, A01) yielded few diatoms, although diatom abundance was generally high, especially in the southern part of the study area where the abundance was generally greater than 20,000 valves/g ( Figure 3 ). The diatom abundance and species diversity in the southern deep-sea basin of the study area were higher than those in the West Caroline Ridge and Yap Trench in the north. Correlation analysis showed a strong positive correlation between diatom abundance and the content of chlorophyll a and POC ( Table 1 ). Silicate was the most significant environmental factor affecting the distribution of diatom abundance, and there was a negative correlation between phosphate and diatom abundance.

In this study, diatoms with relative percentages more than 10% were designated as dominant species, and those with percentages greater than 5% but less than 10% were designated as sub–dominant species (Pokras and Molfino, 1986). Therefore, the dominant species were Azpeitia africanus, Azpeitia nodulifera, Coscinodiscus sp., Hemidiscus cuneiformis, Synedra sp., Synedra tabulata, Thalassiosira pacifica and Thalassiosira leptopus, with the sub-dominant species being Coscinodiscus argus, Coscinodiscus decrescens, Roperia tesselata, Thalassionema nitzschioides and Thalassiosira excentrica ( Table 2 and Figure 4 ). Azpeitia nodulifera (an oceanic warm-water species; Shen et al., 2017a) had the highest mean percentage content, followed by H. cuneiformis (an oceanic warm-water species; Shen et al., 2017a) and T. pacifica (a cosmopolitan species; Halse and Syvertsen, 1996). Overall, these three species together accounted for approximately 80% in this area.

All stations in the study area were dominated by warm-water diatoms ( Figure 5 ), which constituted on average 79.81% of the diatom abundance. The most abundant species were A. nodulifera, H. cuneiformis and A. africanus, which made up 78.62% of the total. Other warm-water species included Asteromphalus heptactis, Asterolampra marylandica, Asterolampra sp., Coscinodiscus radiatus, Mastogloia cocconeiformis, Rhizosolenia bergonii, Roperia tesselata and Triceratium pentacrinus. This diversity of warm-water species was consistent with the environmental conditions of the Western Pacific Warm Pool, where the study area was located. In addition, because of the location in the deep sea, the diatoms were mainly planktonic taxa (average 99.67%) and benthic diatoms occurred only sporadically in the study area ( Figure 5 ).

Ethmodiscus rex fragments were widely distributed in the West Caroline Basin ( Figure 6 ). The valves of E. rex are huge and readily broken, so most of the individuals recorded were present as fragments. High concentrations of E. rex fragments were mainly recorded at stations in the deeper basins and troughs in the southern part of the study area, whereas the number of fragments in the shallow West Caroline Ridge was generally low. The correlation analysis showed that E. rex abundance had a weak positive correlation with WD. The study of E. rex in the southern Mariana Trench by Cai (2019) also found that this species was more abundant in deeper waters and less abundant in shallower waters, and that there was a positive correlation between E. rex abundance and WD, consistent with our findings.

For the statistical analysis of species data, species with relative percentages > 2% were selected and rare species (those occurring at fewer than five stations) were excluded. The environmental data included WD, SST, SSS, silicate, nitrate, phosphate, DO, POC, NPP and chlorophyll a. The RDA results for environmental variables and samples, and for environmental variables and diatom species, are provided in Table 3 and Figures 7 , 8 . The first two RDA axes explained 35.8% and 23.2% of the total variation, respectively. Correlation between environmental variables and RDA axes are shown by both length and angle of arrows. Using cluster analysis, the 80 samples could be divided into four groupings based on the composition of the diatom species and diatom abundance ( Figures 7 , 9 ).

In the study area, warm-water diatoms possessed an absolute advantage, with the lowest percentage abundance of these diatoms being 44.44% at station I06, indicating a tropical oceanic environment. The tropical pelagic species A. nodulifera was the most abundant diatom at each station, consistent with the view that A. nodulifera is the main component of diatom assemblages in the surface sediments of the tropical Pacific (Jousé et al., 1971). E. rex fragments are also widely distributed in this area, corresponding with the nutrient-poor environment in the Western Pacific Warm Pool (Abrantes et al., 2007). Compared with study area I in the West Philippine Basin ( Figure 1 ), there are the same four dominant species, A. africanus, A. nodulifera, H. cuneiformis and T. leptopus, indicating that the two areas possess similar growth conditions for diatoms. However, the proportion of warm-water diatoms, especially A. nodulifera, in the West Caroline Basin was obviously greater than in the West Philippine Basin as a result of latitudinal differences and a more typical pelagic environment.

T. nitzschioides is widely distributed at low latitudes, and is often used as an indicator of upwelling and increased production in coastal areas; however, the species is as abundant in nutrient-poor open oceans as in coastal areas, and is described as a cosmopolitan species (Kobayashi and Takahashi, 2002; Ren et al., 2014; Lyle and Baldauf, 2015). In this study, T. nitzschioides was mainly detected in the high-productivity area in the southern West Caroline Basin ( Figure 10 ). Considerable upwelling (3-10°N, strongest at 6-7°N) is observed in this area (Long et al., 2021), suggesting that T. nitzschioides is also good proxy for upwelling and increased production in the open ocean. Similarly, Kobayashi and Takahashi (2002) found T. nitzschioides may reflect weak upwelling in the equatorial western Pacific.

The distribution of diatoms in sediments is directly related to the different temperature and salinity conditions in the euphotic zone; however, diatom frustules experience dissolution in the water column, current transport and various geological processes during sinking and after deposition on the seafloor. As a result, there are many factors affecting the distribution of diatoms in sediments (Zhang et al., 2013; Astakhov et al., 2015). Using the diatom assemblage distribution ( Figure 9 ), the study area was divided into three regions: West Caroline Basin, West Caroline Ridge and Yap Trench. The West Caroline Basin mainly contains diatom assemblage I, the West Caroline Ridge is dominated by diatom assemblage II, and the Yap Trench has a mixed assemblage and more complex diatom distribution.