Paleoecological evidence for decadal increase in phytoplankton biomass off northwestern Australia in response to climate change

Abstract
 Ocean warming can modify the phytoplankton biomass on decadal scales. Significant increases in sea surface temperature (SST) and rainfall in the northwest of Australia over recent decades are attributed to climate change. Here, we used four biomarker proxies (TEX86 index, long‐chain n‐alkanes, brassicasterol, and dinosterol) to reconstruct approximately 60‐year variations of SST, terrestrial input, and diatom and dinoflagellate biomass in the coastal waters of the remote Kimberley region. The results showed that the most significant increases in SST and terrestrial input occurred since 1997, accompanied by an abrupt increase in diatom and dinoflagellate biomasses. Compared with the results before 1997, the average TEX86H temperature during 1997–2011 increased approximately 1°C, rainfall increased 248.2 mm, brassicasterol and dinosterol contents increased 8.5 and 1.7 times. Principal component analysis indicated that the warming SST played a more important role in the phytoplankton increase than increased rainfall and river discharge.


| INTRODUCTION
Phytoplankton is a key component in marine ecosystems, and its variations in abundance and species composition are sensitively coupled with short-and long-term environmental changes, and consequently influence the structure and function of ecosystems, for example, biogeochemical cycles and the food web (Field, Behrenfeld, Randerson, & Falkowski, 1998). Over the last several decades, phytoplankton regime shifts, for example, the changes in biomass and species composition and shifts between diatom and nondiatom communities, have been widely observed in many coastal ecosystems (Smetacek & Cloern, 2008). Most evidence demonstrated that nutrient enrichment is a principal driving factor for phytoplankton regime shifts in coastal waters (Anderson, Glibert, & Burkholder, 2002), and ocean warming could enhance this process, affecting the distribution and productivity of phytoplankton in the ocean (Irwin, Finkel, Müller-Karger, & Troccoli, 2015). For example, the warmer sea surface temperature (SST) and lower turbidity in the North Sea have increased phytoplankton biomass, even though nutrient concentrations have been decreasing since the 1980s (Mcquatters-Gollop et al., 2007).
Due to limited observational data, it is challenging to distinguish between the impact of climatic variability and anthropogenic disturbances on the phytoplankton in coastal waters. Paleoecological methods, using geochemical and biological information preserved in the sediment to reconstruct the short-or long-term environmental change, have supported significant findings in marine research, although a series of biological, chemical, and physical factors (e.g., water depth, temperature, salinity, grain size, and degradation) during sedimentation and preservation can impact the results (Fischer & Wefer, 1999). A few proxies have been widely applied to reconstruct sea F I G U R E 1 Study location: Cygnet Bay, northwestern Australia (a), water depth (m) in King Sound (b), and sampling site (c: reference site) 115°E temperature, terrestrial input, and phytoplankton biomass, due to their biosynthetic specificity and resistance to degradation in the sediment. Schouten, Hopmans, Schefuß, and Sinninghe Damsté (2002) proposed TEX 86 (TetraEther indeX of tetraethers consisting of 86 carbon atoms) as a proxy for SST, based on the relative distribution of marine archaea isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs). The selected GDGTs are membrane lipids synthesized by Thaumarchaeota, which contain different numbers of cyclopentane and cyclohexane rings.
It has been demonstrated that the addition of rings into GDGTs enables archaea to adjust membrane stability in response to temperature changes (Chong, 2010;Uda, Sugai, Itoh, & Itoh, 2001). The long-chain n-alkanes (C 27 + C 29 + C 31 ), specific to higher land plants, can help to interpret the impact of terrestrial input in marine sediments in terms of changes in rainfall, river discharge, or dust input (Eglinton & Hamilton, 1967;Seki et al., 2003). A few sterols are verified biomarkers for diatoms and dinoflagellates, for example, dinosterol (4α, 23, 24-trimethyl -5α-cholest-22(E)-en-3β-ol) is produced almost exclusively by dinoflagellates, and brassicasterol (24-methylcholest-5, 22(E)-dien-3-ol) is a commonly used diatom biomarker, even though it has been reported in many algal classes (Rampen, Abbas, Schouten, & Sinninghe Damste, 2010;Volkman et al., 1998). The analysis of these compounds in the sediment core can help to reconstruct the long-term changes in the environmental change and phytoplankton community as well as their correlation.
In this study, Cygnet Bay, located in the Kimberley, northwestern Australia ( Figure 1a), was chosen to study the phytoplankton regime shift in response to the climate change. The Kimberley is a remote coastal region with very limited anthropogenic activity and significant climatic variability. During the past several decades, SST and rainfall have been reported as the most prominent climate-induced changes in the northwest coast of Australia. For example, annual averaged SST has warmed by ca. 0.6°C in the past 50 year (Lough, 2008) and annual rainfall increased approximately 50% since 1950 (Feng, Li, & Xu, 2013;Shi et al., 2008). More recently, Furnas and Carpenter (2016) found that primary production in northern Australia increased more than twofold post-1990 compared to the 1960s, but considering the paucity in data, they attributed the difference to the improvements in productivity measurements. Liu, Peng, Keesing, Wang, and Richard (2016) analyzed the variation in organic matter in the sediment cores, which were collected from Cygnet Bay, and they suggested that the variability in climatic signals (rainfall and temperature) might explain the increase in marine organic matter over decadal scales. Therefore, it warrants further examination to elucidate whether the increase in phytoplankton production in the northwest of Australia since the 1990s was related to climate-induced SST and rainfall changes. Diatoms and dinoflagellates are the dominant phytoplankton in Kimberley coastal waters (Thompson & Bonham, 2011) thereby allowing us to use brassicasterol and dinosterol to represent diatom and dinoflagellate biomasses. Four proxies (TEX 86 index, long-chain n-alkanes, brassicasterol, and dinosterol) were chosen and analyzed, using the dated sediment cores from Cygnet Bay, to reconstruct the variations of SST, terrestrial influence, and the diatom and dinoflagellate biomasses, respectively. Validity of biomarker reconstruction and the change in SST and phytoplankton over time are discussed in the context of historical observation data.

| Study area and core information
Cygnet Bay has an area of approximately 150 km 2 and an average water depth of 10 m and is located in King Sound, which is the receiving water body of the Fitzroy River catchment (Figure 1b) (Wolanski & Spagnol, 2003). The cores used in this study were from the collection of Liu et al. (2016) in 2011, by SCUBA divers using a push core with a 6 cm internal diameter. Four replicate cores were taken at each site to enable a range of different parameters to be analyzed. The purpose of the study by Liu et al. (2016) was to investigate the impact of pearl farming on the sediment quality; hence, they chose one contrasting reference site 1.5 km away from the pearl aquaculture area ( Figure 1c; reference site: 16°32′S, 122°59′E; water depth: 9.8 m). In this study, one of the cores (106 cm long) from the reference site was used for biomarker analyses (TEX 86 , long-chain n-alkanes, brassicasterol, and dinosterol). These biomarkers were not used in the previous study of Liu et al. (2016), as they focused on the analysis of chronology and geochemical parameters (organic matter, carbon and nitrogen isotopes, biogenic silica) to examine whether aquaculture had induced any change in the sediment over time. The replicate core used in this study had a similar sediment texture compared to other cores at the reference site, and the median grain size (d 50 ) between replicate cores was significantly correlated at a 95% confidence level (Liu et al., 2016). The core we used covered a

| Chemical analysis of four biomarker proxies
The core used for biomarker analysis was sectioned at 1-cm intervals and freeze dried prior to biomarker analysis. Sample processing and instrumental analyses of biomarker proxies were performed at Ocean University of China, using the same methods described in Li, Zhao, Tian, and Li (2013). Briefly, about 5 g of sediment was extracted four times with dichloromethane/MeOH (3:1, v/v), after adding internal standards (n-C 24 D 50 , C 19 n-alkanol and C 46 GDGT). Extracts were hydrolyzed with 6% KOH in MeOH. The neutral lipids were extracted with hexane and separated into two fractions using silica gel chromatography. The nonpolar lipid (containing n-alkanes) fraction was eluted with hexane, and the polar lipid fraction (containing sterols and GDGTs) was eluted with dichloromethane/methanol (95:5, v/v).
Subsequently, the polar fraction was divided into two parts, one de- The analytical accuracy was better than 0.006 for BIT index in our laboratory.

| Data source and statistical analysis
In order to verify the accuracy of TEX  ERSST during 1940ERSST during -1978ERSST during , 1978ERSST during -1986ERSST during , and 1986ERSST during -2011 were 27.9°C, 28.2°C, and 28.4°C, respectively.
Thus, the increasing trend of SST during 1986-2011 is consistent between the two parameters.

| Long-chain n-alkanes record
The contents of long-chain n-alkanes in marine sediments can help to interpret the variations of terrestrial input. They ranged from 38.  year, higher than during 1963-1996 (2.53 × 10 9 m 3 /year). Thus, the patterns of long-chain n-alkanes, rainfall, and river discharge are broadly consistent in this region, indicating an increase after the late 1990s.

| Principal component analysis (PCA)
PCA of the four proxies showed the first two principal components (PC1 and PC2) were responsible for 82.8% of total variance. PC1 ac-

| Validity of reconstructed proxies
TEX H 86 is suitable for SST reconstruction in tropical or subtropical regions (>15°C; Kim et al., 2012) and has been applied in the Australian region displaying a good linear correlation with instrumental annual SST (Chen, Mohtadi, Schefuß, & Mollenhauer, 2014;Smith et al., 2013). The warming trend of TEX H 86 and ERSST in Cygnet Bay is similar, but there are some discrepancies in the warming rate (Figure 2). This phenomenon was reported in previous studies, for example, northwest Africa (Mcgregor, Dima, Fischer, & Mulitza, 2007), northeast Hong Kong (Kong et al., 2015), and the shelf of Western Australia (Zinke et al., 2015). A few factors could lead to the difference between TEX H 86 temperature and ERSST. One is the errors from the measurement of ERSST as (1) sporadic SST measurements over the historical period can result in large uncertainties in ERSST (Kennedy, 2014;Smith et al., 2008); (2) the 2° × 2° resolution of ERSST could underestimate temperature variability in shallow waters, because the rapid warming rate was more frequently observed in the nearshore than offshore (Lima & Wethey, 2012). Other possible errors are from (2) some studies show the TEX H 86 temperature may be skewed due to seasonality in growth or export of Thaumarchaeota (Jia, Zhang, Chen, Peng, & Zhang, 2012;Leider, Hinrichs, Mollenhauer, & Versteegh, 2010). In general, TEX H 86 captured the warming trend in northwest Australia. Observed air and sea temperatures revealed significant warming since the 1950s and accelerated warming rates since the 1980s (Lough, 2008;Lough & Hobday, 2011). An increase in anomalously warm ocean conditions along the western Australian coastline has been recorded since the late 1990s, strongly influenced by the strengthening of the Indonesian Throughflow in response to increases in Pacific trade winds (Feng et al., 2011. Coral temperature records from northwestern Australia also indicated long-term warming of coastal waters with the highest temperature anomalies recorded during the 1980s-2010 (Zinke et al., 2015). These similar warming patterns therefore confirm the applicability of TEX H 86 index in our study area.
Long-chain n-alkanes have been widely used to evaluate the terrestrial influence on the ocean (Eglinton & Eglinton, 2008). In  Furnas and Carpenter (2016) found primary production increased ca. twofold in 1990-2013 compared to 1960-1962

| Phytoplankton biomass in response to warming SST and terrestrial influence
Temperature, salinity, irradiance, and macronutrient concentrations are regarded as the fundamental environmental factors determining phytoplankton niches. Recently, statistical analysis of time series data in the Baltic Sea and Caribbean Sea showed that the importance of temperature, salinity, and irradiance for the niches of diatoms and dinoflagellates is even higher than macronutrient concentrations (Gasiūnaitė et al., 2005;Irwin, Nelles, & Finkel, 2012;Mutshinda, Finkel, & Irwin, 2013). In this study, a significant concurrent shift occurred in 1997. Compared with the average values of four proxies before 1997, the average TEX H 86 temperature during 1997-2011 increased approximately 1°C, rainfall increased 248.2 mm, and brassicasterol and dinosterol sediment contents increased 8.5 and 1.7 times, respectively. The PCA indicated that warming temperature has a more significant impact on the increases in diatom and dinoflagellate biomass than terrestrial input.
The driving mechanism of warming temperature on phytoplankton biomass is complicated as most results from open sea showed that ocean warming can enhance the vertical stratification, which could reduce the nutrient supply to the mixed-layer and consequently reduce phytoplankton biomass in the surface water (Behrenfeld et al., 2006;Richardson & Schoeman, 2004). However, some studies indicated that warming temperature can accelerate nutrient recycling by bacteria, resulting in phytoplankton increase (Taucher & Oschlies, 2011). In Cygnet Bay, the seawater is well mixed due to the shallow depth (9.8 m) and strong tidal action all year round.
Thus, the increase in phytoplankton biomass was more possibly attributed to positive physiological action and fast nutrient turnover. Some studies in northwestern Australia emphasized a mechanism of ocean-coast interaction that could influence the coastal phytoplankton biomass, for example, in the North West Cape, the increase of phytoplankton biomass in coastal waters could be caused by increased nutrients transport from the deep sea through enhanced upwelling (Furnas, 2007). In Darwin Harbour, Burford, Alongi, Mckinnon, and Trott (2008) found that the oceanic inputs of nutrients to the estuary were a primary contributor. However, due to a lack of observational data, the mechanism of the ocean-coast interaction in the Kimberley needs further exploration.
The sediment core analysis revealed not only the increased phytoplankton biomass since 1997 but also the different shifting pattern between diatoms and dinoflagellates. Dinoflagellates showed an earlier but slower increasing trend compared to diatoms (Figure 4). Thompson and Bonham (2011) analyzed the phytoplankton communities during a 2010 research voyage in the Kimberley and found the pigment proportion of diatoms in shallow water (<50 m) was much higher than for dinoflagellates. This supports our result in the upper section of the core. In general, diatoms are favored in niches with higher macronutrients, higher turbulence, lower salinity, and lower temperature than the dinoflagellates, and thus, diatoms often dominate in the coastal and estuarine waters (Berdalet, 1997;Margalef, 1978). Silicate is a critical, limiting macronutrient for diatom growth but not for dinoflagellates, and the source of silicate in coastal waters is mainly through riverine input. Thus, increased riverine input and rainfall often enhance the supply of silicate and decrease the salinity which can provide more suitable conditions for diatom growth. In this study, diatom and dinoflagellate biomarkers did not display any correlation with long-chain n-alkane (proxy for terrestrial inputs); however, the increased river discharge and rainfall after 1997 provide a mechanism by which conditions after that time favored diatoms over dinoflagellates.
In summary, the paleoecological evidence from Cygnet Bay demonstrated that SST and terrestrial input have significantly increased since 1997 in the Kimberley region, and the biomasses of diatoms and dinoflagellates corresponded to these changes with an increasing trend. In comparison, warming SST played a more important role for the phytoplankton increase than increased rainfall and river discharge.

ACKNOWLEDGMENTS
The study was funded by the National Natural Science Foundation

CONFLICT OF INTEREST
None declared.

AUTHORS' CONTRIBUTION
Zineng Yuan contributed to interpretation of data and drafting the work; Dongyan Liu contributed to the conception of the work, revising the draft and final approval of the version to be published; John Keesing contributed to design of the work and acquisition of samples; Meixun Zhao contributed to revising the draft and analysis methods; Shixin Guo contributed to the sample analysis; Yajun Peng contributed to interpretation of data; Hailong Zhang contributed to the sample analysis.