The sub-fossil diatom distribution in the Beibu Gulf (northwest South China Sea) and related environmental interpretation

Study area description

Geomorphology: The Beibu Gulf is a semi-closed fan-shaped continental shelf shallow sea in northwest SCS (Dong et al., 2019; Fig. 1). The water depth generally increased from northwest coastal areas towards the southeastward shelf break at about 200 m, but is less than 100 m in the middle and south of this gulf. However, water depth in broad areas in the northern part (north of 20°N) does not exceed 50 m. The middle Beibu Gulf has a relatively flat sea floor, while depths exceed 100 m in the southern part and near shelf break.

Climate: The research area has a strong monsoon climate. The winter is characterized by northeast monsoon, whereas during summer the southwest monsoon prevails with northward winds, and accompanying southeast wind. The northeast monsoon temporal interval is longer than the southwest one. The total annual rainfall is about 1,100–1,700 mm, and peak rainfall lasts from May to September (the rainy season) with an average monthly rainfall over 100 mm (Li et al., 2017).

Temperature and Salinity: Average monthly sea surface temperature of Beibu Gulf from 1994 to 2013 oscillated between 20.3–29.9 °C as derived from satellite data (Ya, Gao & Dong, 2015). In general, the spatial pattern of temperature distribution is cold water in the north and warm water in the south (Chen et al., 2009; Ya, Gao & Dong, 2015). The water temperature gradient between north and south usually amounts to ca. 5 °C, not exceeding 2 °C during the summer due to strong solar radiation over the entire SCS surface (Chen et al., 2009).

Due to higher fresh water runoff in the north and northwest, the northern Beibu Gulf has lower sea surface water salinities. On the contrary, the southern region is characterized by higher salinities that result from sea-water intrusion from SCS (Chen et al., 2009; Ya, Gao & Dong, 2015). The salinity displays seasonal variation with lower summer values due to fresh water discharge of ca. 21.2 psu in the coastal area. Winter salinity of coastal waters amounts to ca. 30.3 psu. In general, coastal area salinity ranges from 26 to 34 psu, but usually exceeds ca. 28 psu (Chen et al., 2009; Li & Li, 1993).

Water circulation: Water circulation in Beibu Gulf is principally driven by wind speed and direction, sea-water density distribution, external open sea surface water mass and fresh water discharge, with the latter factor playing important role in long-term sea-water transportation (Wang et al., 2018). The northeast monsoon during winter induces a sinistral circulation, however, lengthy controversy continues over the summer circulation direction and mechanisms (Gao, Wu & Ya, 2017; Xia, Li & Shi, 2001). Summer circulation appears to follow a process where southeast monsoon induces a dextral circulation in Beibu Gulf, while the density gradient induces sinistral flow, with density currents being stronger than wind-driven, especially in the surface layer (Chen et al., 2009; Xia, Li & Shi, 2001; Su & Yuan, 2005). Also, westward flow through the Qiongzhou Strait plays an important role in formation of summer sinistral flow (Chen et al., 2019). In addition, coastal current flow from northern SCS along the southern Hainan Island coast plays an important role in Beibu Gulf circulation, particularly in winter to spring seasons (see Fig. 2).

Laboratory measurement

Sixty-six surface box core sediment samples were collected from eastern Beibu Gulf (Fig. 1) during geological cruises organized by Guangzhou Marine Geological Survey, China during 2007 to 2012. We selected the core top 1st cm to analyze sub-fossil diatom assemblages, by permission from GMGS-CGS (No. GZH201500207).

Diatom analysis: Ten grams of dry sediment sample were processed at Xiamen University (China). Twenty ml of 30% hydrogen peroxide was used to oxidize organic matter, followed by 10 ml of 10% hydrochloric acid added to remove carbonates. Sieves (10 µm mesh size) were used to remove the large quantities of clay particles, and thereafter centrifuged twice with heavy ZnI₂ liquid (density 2.4 g/cm³). Thirty µl of vigorously shaken cleaned diatom suspension was pipetted onto a 22 × 22 mm cover slip and left to dry at room temperature. The cover slips were transferred onto labeled slides, mounted with Naphrax and heated to remove air bubbles and humidity (Zhang et al., 2016). Three slides were prepared for each sample. At least 300 diatom valves were counted at the species or genus level in a slide made for each sample with a microscope magnification of ×400 and ×1000 (oil immersion).

Relative and absolute abundance of sub-fossil diatoms was estimated on the basis of diatom valve counts in particular slides. Diatom absolute abundance (=DAA) was calculated with the counted valves / weight of dry sediment used for sample processing. The diatom relative abundance (=DRA) was expressed as percentage of each diatom species/divided by total observed diatom abundance in each sample. The identification follows (Jin, Cheng & Lin, 1982; Jin, Cheng & Liu, 1992; Lan, Chen & Liu, 1995; Witkowski, Lange-Bertalot & Metzeltin, 2000; Guo & Qian, 2003). Separately, the freshwater diatom percentage, total percentage of warm water and higher salinity taxa, and planktonic to benthic (P/B) diatom ratio were calculated.

Environmental data set: The water depth of sampling site was recorded. The physical parameters of annual mean sea surface temperature (SST), annual mean sea surface salinity (SSS) and trophic state parameter of annual mean chlorophyll a concentration (Chl. a) data were abstracted through satellite remote sensing data set from the European Space Agency (European Space Agency, 2019). These four parameters were sorted as environmental variables in this study.

Multivariate analysis: In order to reduce the complex signal of sub-fossil assemblages, cluster analysis was used where diatom taxa with DRA in one individual site exceeded 5%. Statistically constrained cluster analysis of diatom species as groups was operated with PC-ORD version 5.0 (MjM Software Design). In this software, the Q-mode and R-mode hierarchical analyses are only based on diatom taxa in each site, the DRA as percentage was treated as a variable, utilizing Sorensen distance and Ward’s merging criterion.

Through correspondence analysis (CA) via the software Canoco 5.1 (Šmilauer & Lepš, 2014), the lengths of gradient (at 2.4 SD units) were obtained. The data were standardized in this software through log-transformation. Due to this gradient value, principal component analysis (PCA) was operated by unconstrained ordination suitably, in order to understand the relationship of diatom species matrix and environmental variables. The 2.4 SD units made redundancy analysis (RDA) valuable to reduce dimensions and obtain eigenvalue, following in constrained ordination model, for understanding the relationship of diatom species matrix and environmental variables.

Diatom abundances

The DAA in studied sites varied within a broad range from 0.22 to 28.0 × 10³ valves/g, with average 5.53 × 10³ valves/g. DAA was higher in the middle and northern parts of inner Beibu Gulf in coastal area and Leizhou Peninsula (Fig. 3). The lower DAA values (below 1 ×10³ valves/g) were observed in two southernmost sites close to continental shelf break and four sites dispersed within middle Beibu Gulf.

Diatom species

The surface sediment samples contained sub-fossil diatom assemblages usually represented by fairly high number of taxa. In total, 218 taxa (species) were identified with microscopic analysis based on above listed references. The identified diatoms with broad ecotypes include freshwater, marine-brackish water (coastal and neritic), warm water and tropical forms. The freshwater species Planothidium hauckianum and Cymbella sp. were identified in two sites (B10 and B27, Figs. 4, 5). Coastal benthic species in marine-brackish water forms widely occurred in this gulf and included Cyclotella striata, C. stylorum, Diploneis crabro, Paralia sulcata, Trachyneis antillarum, T. aspera and Triceratium favus.

A group of warm water species had wide distribution in the study area and included Bacteriastrum hyalinum, Campylodiscus brightwellii, Ditylum brightwellii, Planktoniella blanda, Rhizosolenia bergonii, Stephanopyxis weyprechtii, Thalassiosira bipartita, T. leptopus, and Trieres chinensis. The percentage of these warm water species ranged from 2.21% to 33.33% (average 15.52%, Fig. 4). The warm water group percentages were higher in middle and south Beibu Gulf, and similarly far away from the coastal zone (Fig. 5).

In the topmost sediment (highest 1 cm) also occurred numerous diatom species considered characteristic for tropical waters of higher salinity (e.g., Asterolampra vanheurckii, Asteromphalus flabellatus, Azpeitia africana, A. neocrenulata, A. nodulifera, Hemidiscus cuneiformis). However, these species occurred with low DRA ranging from 0.63% to 17.15% (average 5.15%, Fig. 4). The tropical diatom forms achieved higher percentages in deep waters of southern and middle northern part of inner Beibu Gulf (Fig. 5).

The diatom P/B ratio ranged from 0.11 to 1.11 (average value 0.39). The diatom P/B ratio was below 1.0 in most sites, but showed a distinct increase in coastal areas, especially in shallow waters less than ca. 25 m water depth in northern Beibu Gulf. And increased P/B values occurred in middle Beibu Gulf over ca. 60 m water depth, and especially in southern Beibu Gulf where values exceeded 1.0 in three sites with deep waters. On the contrary, the P/B value show an amplitude function and without legible trend in areas where water depth between ca. 30 m to ca. 60 m (Figs. 4, 6).

Similarly to Pokras & Molfino (1986), we selected the diatoms species with relative abundance >5% in each sample which were considered as dominants. For the study area, 20 such species have been designated (Fig. 4, Table 1).

Statistical analyses results

Q-mode statistical analysis

To analyze the relationships among these studied sites, a Q-mode hierarchical cluster analysis was conducted (Fig. 6). The resulting dendrogram represents a grouping of sites according to species’ DRA, and two major clusters, Clusters 1 and 2 are recognized. Cluster 1 can be further subdivided into three subclusters, 1A, 1B and 1C (Fig. 6). In order to clearly express these dominant diatom species in the Q-mode hierarchical cluster, this article sorts diatom species’ average DRA over 10% in one sub-cluster as major dominant diatom species, and species with average DRA less than 10% are chosen as minor dominant diatoms.

Subcluster 1A comprised six sites, which are in middle Beibu Gulf located from north to south with water depths that exceed 50 m (Fig. 7), and mainly dominated by Campylodiscus brightwellii and Cyclotella striata, with increased percentages of Paralia sulcata and Planktoniella blanda (Table 2). However, three other diatom species occurred with lower abundances, Actinocyclus octonarius, Azpeitia nodulifera, Coscinodiscus argus in this subcluster.

Table 2:

The most abundant diatom species in the four subclusters.

The species are ranked according to decreasing average DRA over 1.5% in each subcluster.

Subcluster	Dominant diatom species
	Major dominant species (average DRA < 10%)	Minor dominant species (average DRA < 10%)
Subcluster 1A		Paralia sulcata (8.59%)
	Campylodiscus brightwellii (14.39%)	Planktoniella blanda (8.30%)
	Cyclotella striata (12.49%)	Actinocyclus octonarius (4.20%)
		Azpeitia nodulifera (3.35%)
		Coscinodiscus argus (1.67%)
Subcluster 1B	Cyclotella striata (22.23%)	Campylodiscus brightwellii (3.72%)
	Paralia sulcata (18.96%)	Actinocyclus octonarius (1.83%)
	Cyclotella stylorum (11.60%)
Subcluster 1C	Cyclotella striata (27.06%)	Paralia sulcata (8.90%)
		Cyclotella stylorum (5.75%)
		Campylodiscus brightwellii (5.57%)
		Azpeitia nodulifera (3.35%)
		Thalassiosira bipartita (2.09%)
		Stephanopyxis weyprechtii (1.64%)
Cluster 2	Paralia sulcata (30.93%)	Cyclotella stylorum (7.00%)
	Cyclotella striata (15.62%)	Azpeitia nodulifera (2.19%)
	Cyclotella stylorum (7.00%)	Podosira stelligera (1.56%)

DOI: 10.7717/peerj.13115/table-2

Subcluster 1B comprised fifteen sites, mainly located in northern Beibu Gulf, with only one site (B56) located southwest of Hainan Island near Sanya City (Fig. 7). In this subcluster, the major dominant diatom species were C. striata, Cyclotella stylorum and P. sulcata, with Campylodiscus brightwellii and A. octonarius as minor dominant diatom species.

Subcluster 1C comprised twenty-five sites occur mainly in middle and southernmost Beibu Gulf and along the west coast of Hainan Island (Fig. 7). Cyclotella striata is the only major dominant species in this subcluster, but other species (e.g., C. brightwellii, C. stylorum, P. sulcata and Thalassiosira bipartita) with higher average DRA graded as minor dominant species. Thus, this subcluster gathered various dominant diatom species, which suggests these to be in a strongly mixed area.

Cluster 2 comprised twenty sites, mainly located in northernmost Beibu Gulf, west of Leizhou Peninsula, and southwestern coast near Sanya City (Fig. 7). P. sulcata and C. striata were major dominant species in this cluster, except for site B59 with A. nodulifera over DRA 13% as most dominant species. In this cluster, C. stylorum also has relatively higher percentages, usually over 5%.

PCA and RDA ordination analysis

The PCA analysis of diatom species showed Axis 1 to Axis 4 with eigenvalues 0.4667, 0.2112, 0.0968, 0.0677, respectively (see Table 3). This explained 84.24% of (cumulative) variation. The PCA analysis results can make clearly the ordination of dominant diatom species in the distribution pattern. C yclotella striata, C. stylorum and Paralia sulcata were the most significant species as illustrated in Fig. 8A, while other diatom species were of lower weight in the matrix. The studied site plot in the PCA analysis (Fig. 8B) showed an order distribution that corresponds to the results of Q-mode hierarchical cluster analysis.

Table 3:

The PCA analysis results for the first four axes.

		Axis 1	Axis 2	Axis 3	Axis 4
Analysis type	Unconstrained
Method	PCA
Total variation	23672.702
Statistic Eigenvalues		0.4667	0.2112	0.0968	0.0677
Explained variation (cumulative)		46.67	67.79	77.47	84.24

DOI: 10.7717/peerj.13115/table-3

Subcluster 1A gathers in corner of the third quadrant, while subcluster 1B gathers in center of the graph (Fig. 8B). Subcluster 1C mainly gathers in the fourth quadrant. Cluster 2 crossed the first and second quadrant. Sites in Cluster 2 along the PCA coordinate axes (Fig. 8B) with higher abundance of P. sulcata.

The RDA analysis between diatom species and environmental variables showed these four explanatory variables to account for 38.14% of variation (Table 4, Fig. 9). The statistic axes 1 to 4 have eigenvalues of 0.2985, 0.0507, 0.0229, 0.0093, respectively (see Table 4). These correspond to the species distribution (A. nodulifera, P. blanda, S. weyprechtii), attributed to deeper-water, higher temperature and salinity sites in middle and south Beibu Gulf. The Chl. a mainly correspond to C. striata and P. sulcata (two top species of higher average abundances in this study, Table 1) which occur mostly in north and southeastern Beibu Gulf.

Table 4:

The RDA analysis results for the first four axes.

		Axis 1	Axis 2	Axis 3	Axis 4
Analysis type	Constrained-P
Method	RDA
Total variation	23672.702
explanatory variables account	38.14%
Statistic Eigenvalues		0.2985	0.0507	0.0229	0.0093
Explained variation (cumulative)		29.85	34.92	37.2	38.14
Pseudo-canonical correlation		0.8131	0.5935	0.47	0.3595
Explained fitted variation (cumulative)		78.27	91.56	97.56	100
Permutation Test Results	pseudo-F = 9.4, P = 0.002

DOI: 10.7717/peerj.13115/table-4

Cluster’s attribution explanation

On the clusters level, Cluster 1 and Cluster 2 occupied about 2/3 and 1/3 of sites in the studied area, respectively. This ratio indicates that major factors largely impact the shallow gulf, although the two clusters are dominated by different diatom species and ordination linked with DRA.

Subcluster 1A formation

Distribution of subcluster 1A is distinctness to align in middle Beibu Gulf from ca. 50 to 90 m water depth. The main dominant species is the marine benthic diatom C. brightwellii, associated with warm water (Jin, Cheng & Lin, 1982; Jin, Cheng & Liu, 1992), and A. nodulifera as minor dominant species indicative of open sea higher temperature and salinity (Winter, 2001). While, the increased relative abundance of C. striata, C. stylorum and P. sulcata coastal species as the main PCA plotted variations (Witkowski, Lange-Bertalot & Metzeltin, 2000; Guo & Qian, 2003; Zong, 1997) indicates impact of coastal currents in middle Beibu Gulf (see Figs. 3, 4 and 8).

Subcluster 1B formation

Subcluster 1B is observed in northern Beibu Gulf with water depth decreasing to ca. 40 m, except for one site (B56) at ca. 90 m water depth in south Beibu Gulf. Compared to subcluster 1A, 1C and Cluster 2, this subcluster is likely to occur in a transition area, bounded to south by subcluster 1A and 1C, and northward linked to Cluster 2. This directly explains formation related to oceanographic conditions in terms of coastal water impact and currents driven from Qiongzhou Strait (Fig. 2), mixing with warm water from SCS. The dominant species C. striata, C. stylorum and P. sulcata occupied most part of this area, as typical coastal species, that implied coastal environment importance in subcluster origin. Common occurrence of C. brightwellii (associated with warm conditions) and A. octonarius (marine planktonic species, Witkowski, Lange-Bertalot & Metzeltin, 2000; Guo & Qian, 2003) in this shallower area shown as minor dominant species in Table 2, provides an effective proof that this area was impacted by other water masses.

Subcluster 1C formation

Subcluster 1C included the largest set of sampling sites in Beibu Gulf with large spatial distribution but divided into two areas (Fig. 7). Comparing to Subcluster 1B, sites positioned in middle Beibu Gulf have a mixed regime with diverse diatom species in terms of their requirements. The typical coastal species C. striata is the only major dominant species in this subcluster to indicate strong impact of coastal waters, but there are other species with average DRA ranging from 5% to 10% (Table 2). Warmer water and higher salinity species occurred, including A. nodulifera, S. weyprechtii and T. bipartita, which indicate warm water from open SCS invading into middle and north Beibu Gulf.

An interesting observation has been made regarding the highest DAA in this area (Fig. 3) which occurred in middle Beibu Gulf (site B45). This station is located within the Red River discharge plume, which can be source to increased diatom discharge and deposition.

In the southernmost part of studied area (sites B66 and B65) near shelf break (Fig. 1), the lowest DAA (Fig. 3) was collaborative with dominance of coastal species and C. striata in particular. Although increased temperature and salinity conditions proximity of the open SCS water mass, these stations would not rather display higher DRA of tropical water species. The explanation can be either a strong downslope current that results from a high sediment surface gradient or occurrence of reworked sediments.

Cluster 2 formation

Cluster 2 appeared in two relatively independent areas in the northernmost and south parts of Beibu Gulf (Fig. 7). The highest average DRA of Paralia sulcata differentiates this cluster from above subclustered sites. The northernmost Beibu Gulf received siliceous acid by runoff (Wang et al., 2015), it provides optimum conditions for brackish water diatom growth and form large depositional flux in these cluster sites (Table 2). In line with this interpretation is the presence of freshwater forms which have been observed at B10 and B27 in the northern coastal area. Meanwhile, north SCS coastal currents transferred coastal water into Beibu Gulf via the Qiongzhou Strait, which increased presence of brackish/coastal species (e.g., C. striata and P. sulcata).

Azpeitia nodulifera is the most dominant species in one site (B59), in southern Beibu Gulf, which suggests strong impact by warm and high salinity water from SCS. While P. sulcata and C. striata were abundant (Table 2) in a few sites in south Beibu Gulf, these were likely transported from inner gulf by the coastal current southeastern of Hainan Island.

Role of critical environmental factors

Water depth

Although a subordinate environmental variable when compared to SSS, SST and Chl. a in this study area (Fig. 9), water depth appears to be one of the critical factors that controlled sub-fossil diatom distribution. As revealed by cluster analyses (Figs. 3, 6, 7), the subcluster 1B sites were grouped in northern Beibu Gulf in relatively shallow water (below 50 m). Subcluster 1A is arranged along the ca. 50 to 90 m water depth contours. In Cluster 2, the sites mainly originate in the northern coastal area with water depths below 25 m. Dominant in all of these sites, but especially in northern shallow water areas is P. sulcata, a typical coastal species. This observation differs from Jiang (1987), who commented that sub-fossil P. sulcata mainly followed the water depth increase to enhance relative abundance on the East China Sea shelf. While, in the western Taiwan Strait, this species has higher abundance in surface sediments of less than 50 m water depth (Zhang, 2009). This may show P. sulcata to be a regional specialist species with preferred water depth that varies with region. In addition, P. sulcata dominated a few sites of Cluster 2 located in southern Beibu Gulf deep water. This is probably related to other factors than water depth, such as trophic state (as revealed by Chl. a in Fig. 9) and lateral current transports as discussed in above context.

In middle Beibu Gulf, especially in areas where water depth exceeds 25 m, diatom assemblage is composed of planktonic forms (cf. subcluster 1C shown in Fig. 7) typical for sedimentation in calm conditions. However, that this assemblage accommodates diatoms from various types of habitats (benthic and planktonic) may imply influence by middle Beibu Gulf currents that transported diatoms to ca. 90 m depths (Fig. 2). The steeper bottom gradient with coarse sediments substrate type, in the southernmost part of this study area to over ca. 130 m depths close to shelf break (Fig. 1), created the anomalous mixture of diatom assemblage in subcluster 1C.

The higher diatoms P/B ratio were mainly observed in southern Beibu Gulf, and also occur in the middle area with deeper water (Figs. 4, 10). The lower ratio may be limited mainly to coastal areas with lower water depths. This P/B ratio is directly related to water depth in the studied area. However, differences from this general pattern were observed in the southernmost sites B65 and B66 with relatively deep water, which revealed lower P/B ratio due to abundant occurrence of coastal species. Usually, the planktonic diatoms inhabit water masses with higher depths, whereas benthic diatoms mainly grow in shallow water photic conditions. Numerous case studies have used the P/B ratio to qualitatively infer water-level changes in lakes (e.g., in Africa, Ekblom & Stabell, 2008; in Asia, Wang et al., 2013), while in the marine system literature this index is less popular. This study may expose water depth constraining diatoms P/B ratio in a broad sense.

Statistical significance for modern and paleoenvironment understanding

Forward the previous quality level studies on sub-fossil diatom in Beibu Gulf (Wang, Jiang & Zhang, 1990; Zhang, Chen & Zhang, 2013), this statistical analysis provides clear information about the sub-fossil diatom distribution pattern in this gulf, which has a complicated environment in terms of oceanographic proxies and hundreds of diatom taxa. The nutrients, seawater salinity, temperature, water depth, water mass currents, and river runoff variables may deeply impact the diatom distribution in surface sediments from the studied area. The Chl. a, SST, SSS and water depth, explain 38.14% of environmental information to impact the diatom species distribution in this gulf. This indicates that other locally significance parameters also impacted the sub-fossil diatom species distribution in this study area.

Except for seawater salinity and temperature constraints on diatom distribution among the subclusters, discussion in connection with water depth indicates that different subclusters have significance for modern environment management based on statistical analysis and understanding. In particular, the ca. 25 m water depth is a lower boundary to northernmost sites in Cluster 2, creating a relative biogeographical area, of potential value for regional environmental protection management in inner Beibu Gulf. This result provides a potential candidate/selection index for a marine ecological red-line definition for sustainable development, hence can be of some help to marine and coastal zone management.

The relationship is perspicuity in qualitative description via this statistical analysis, which contributes to a useful direction for further paleoceanographic and paleoenvironmental study in this region. Furthermore, given combined measurement oceanographic parameters, the quantitative analysis is valuable for statistical relation interpretation to sub-fossil distribution in Beibu Gulf.