CHANGES OF THERMOCLINE DEPTH AT THE SUMBA ISLAND OFFSHORE BASED ON PLANKTONIC FORAMINIFERAL ASSEMBLAGES AND ITS IMPLICATION TO EUTROPHICATION SINCE THE LAST DEGLACIATION (~18 ka BP): A PRELIMINARY STUDY

Changes of the thermocline depth (DOT) at the Sumba Island offshore are not well-known compared to the DOT changes in the Timor Sea, the main exit passage of the Indonesian Through-flow (ITF). Planktonic foraminiferal assemblages in cores collected from the southwest Sumba offshore (ST08) and Sumba Strait (ST12, ST13, and ST14) were used as a tool to infer the DOT and paleoproductivity changes at the Sumba Island offshore. The DOT changes were indicated from the thermocline and mixed layer dwellers’ relative abundance while the paleoproductivity changes were indicated from the relative abundance of Neogloboquadrina dutertrei. This study suggests a contrast between the DOT pattern at the Sumba Island offshore and the DOT pattern in the Timor Sea during the Last Deglaciation–Holocene. The contrast DOT pattern indicated that the multi-millennial changes of the Australian-Indonesian monsoon (AIM) during the Last Deglaciation– Holocene were the main factors behind the DOT changes in this region while the effects of El Niño Southern Oscillation (ENSO) –like, Indian Ocean Dipole (IOD) –like, and ITF to DOT changes were minimal. Paleoproductivity enhancement at the Sumba Island offshore was not solely related to the monsoon-driven coastal upwelling intensification, which resulted in the DOT shoaling and eutrophic condition. The increase of nutrient availability in surface water due to river runoff increase and changes in the lifted water mass nature were also able to enhance productivity in this region.


Introduction
The Depth of Thermocline (DOT), which is the distance between the upper limit of the thermocline layer and the ocean surface (Lana et al., 2017), is one of the most studied parameters in paleoceanographic studies (Spooner et al., 2005;Ding et al., 2013;Kwiatkowski et al., 2015). The study of the DOT changes is important due to their association with the variation of upwelling intensity and marine productivity (Brasier, 1995;Müller and Opdyke, 2000;Holbourn et al., 2005). The increase of the upwelling intensity is indicated in the shoaling of the DOT which triggers the eutrophication process as the nutrient-rich cool water layer reaches the photic zone (Brasier, 1995;Susanto et al., 2001). This condition is also known as eutrophic, which is associated with the regime of higher marine productivity (Brasier, 1995 On the other hand, the condition when the nutrient-rich cool water layer does not reach the photic zone is known as oligotrophic, associated with the regime of lower marine productivity (Brasier, 1995;Spooner et al., 2005;Andruleit et al., 2008).
The DOT in the Indonesian region varies spatially with the shallower DOT in its western part and the deeper DOT in its eastern part (Lana et al., 2017). The shallower DOT in western Indonesia is associated with the development of coastal upwelling at the south Sumatra offshore-Lesser Sunda Islands, also known as the Java upwelling region, especially during the Australia-Indonesian winter monsoon (AIWM) (Susanto et al., 2001(Susanto et al., , 2006Andruleit et al., 2008). In eastern Indonesia, the upwelling-related DOT shoaling is hindered by the maximum flow of ITF's surface water, and as a result, the DOT is relatively deeper than in western Indonesia The Sumba Island offshore is situated at the confluence of the ITF's exit passage and the Java upwelling region. The mechanism for the past DOT and paleoproductivity changes is still not well understood in this region, as most of the previous studies focused on the main ITF exit passage (Timor Sea) ( This preliminary research aimed to reveal the mechanism of paleoproductivity changes at the Sumba Island offshore, were they accompanied by the DOT changes? In this research, the DOT and paleoproductivity changes were inferred from foraminiferal assemblages of three marine sediment (gravity) cores taken from Sumba Strait (ST12, ST13, and ST14) and a gravity core taken from southwest Sumba offshore (ST08) (see Figure 1).

Climate and Oceanographic Setting
The oceanography of the Indonesian seas is highly in-  Figure 2). During the AISM, the northwest winds bring the moisture-rich air to the Indonesia region and induce the wet season, while the drier southeast winds during AIWM deliver the dry season (Wheeler and McBride, 2005). The rainfall difference between the wet and dry seasons in the Indonesian region is most prominent in its southern part (from Southern Sumatra to the Lesser Sunda Islands) which indicates the stronger monsoon influence (Aldrian and Susanto, 2003; Mohtadi et al., 2007).  Figure 2). This water mass blocks the ITF surface water and causes the formation of thermocline-dominated ITF, which in turn causes shoaling of the DOT in the ITF exit passage (Gordon et al., 2003;Qu et al., 2005). The northwest winds also affect the direction of another surface current, i.e. South Java Current (SJC), which flows to the southeast, bringing the water mass from the equatorial Indian Ocean to the Timor Sea before merging with the Leeuwin Current (LC), and the water mass from ITF to form the South Equatorial Current (SEC) (Schott and McCreary, 2001

Material and Methods
Four gravity cores collected during the 2016 Widya Nusantara Expedition (E-WIN) from off southwest Sumba (ST08) and from Sumba Strait (ST12, ST13, and ST14) were used as the research materials (see Table 1). The samples for foraminifera determination and count- , and Putra and Nugroho (2020). A total of 250 sub-samples were collected and ~5 g from each sub-sample were prepared using the swirling method (without H 2 O 2 ) to disengage the foraminifera specimens from the mud. Each sample was washed on the top of 200 mesh (0.074 mm) sieve and the residue was dried at the temperature of 60 o C for ~6 hours. Quantitative determination of foraminiferal taxa was preceded by splitting the prepared sample until ~300 specimens of foraminifera were present in each part. All identified taxa in one part were counted while new taxa detected in the remaining parts were counted as one specimen even after normalization. The foraminiferal counts were normalized against the number of splits and the weight of the samples (Ardi et al., 2019). Planktonic foraminifera descriptions by Kenneth and Srinivasan (1983) and Bolli and Saunders (1985) were referred. The sample preparations and the determination and counting of foraminifera were conducted at the Sedimentary Laboratory of the Research Center for Geotechnology of the Indonesian Institute of Science (LIPI) and the Micropaleontology Laboratory of Institut Teknologi Bandung (ITB).
Accelerator Mass Spectrometer-measured radiocarbon (AMS 14 C) ages were only available on five depth intervals of core ST08 (see Table 2) (Ardi, 2018). The radiocarbon ages were converted to calibrated calendar and radiocarbon ages with High Probability Density Range (HPD) Method (Bronk Ramsey, 2009) (for 24-25 cm, 74-75 cm, and 104-105 cm depth intervals), and the intercept of radiocarbon age with calibration curve (Talma and Vogel, 1993) (for 166-167 cm and 235-236 cm depth intervals) based on the MARINE13 calibration datasets (Reimer et al., 2013). The age model for the core ST08 was generated using the Clam package (version 2.3.4) on R (Blaauw, 2010, 2020; R Core Team, 2013). This standard statistic approach is the better choice to rapidly and systematically produce age-models for the low-resolution dating sites since the use of complicated Bayesian age-modeling methods might not add much accuracy and precision (Blaauw, 2010). Twenty runs were done, and a model with the lowest goodness-of-fit (-log) was chosen. This model employed linear interpolation, weighted average-based calendar age point estimates for depths, and the standard Gaussian distribution. The core top sediment was assumed to be aged -66 Before Present (BP) (Present=1950 Anno Domini/AD, core ST08 was retrieved in 2016) and the sedimentation rates between the aged intervals were assumed to be constant (see Figure 3 and Table 3). The sedimentation rates abruptly increased in 25-74 and 75-104 cm depth intervals which were coeval to Mid Holocene, while the sedimentation rates of 0-24 cm, 105-166 cm, and 167-236 cm depth intervals (coeval to Late Holocene, Late Deglaciation-Early Holocene, and Early Deglaciation) were relatively lower (see Figure  3). Relative age based on planktonic foraminiferal zonation (Bolli and Saunders, 1985) was also utilized to establish the chronological framework in this study. Pleistocene-aged sediments were only inferred on core ST08, ST13, and ST14 while Holocene-aged sediments were inferred on all cores, including core ST12.
The relative abundances of thermocline dwellers (i.e. Pulleniatina obliqueloculata, Neogloboquadrina spp. and Globorotalia spp.) and mixed layer dwellers (i.e. Globigerinoides ruber, Globigerinoides trilobus, and other Globigerinoides taxa) planktonic foraminifera (ratio of the total number of thermocline/mixed layer dwellers against the total number of planktonic foraminifera in a subsample) (see Supplementary Materials) were used to infer the changes in the DOT, as suggested by Bé et al. (1977) and Ravelo et al. (1990). The shoaling (deepening) of the DOT was indicated by the increase

Results and Discussion
The proxies used in this study (thermocline dwellers, mixed layer dwellers, and N. dutertrei) exhibit changes within the Holocene and Pleistocene (Last Deglaciation) periods. Spatial differences of thermocline dwellers, mix layer dwellers, and N. dutertrei changes are also observed in the research area.

Core ST08 (off southwest Sumba)
Based on the age model of core ST08 (see Figure 3), the first occurrence (FO) of Globorotalia (Gl.) fimbriata which indicates the beginning of the Holocene is coeval to ~11 ka BP (see Figure 4). The abrupt changes in thermocline dwellers, shallow dwellers, and N. dutertrei were observed around this time, which indicated a contrast abundance pattern during the Last Deglaciation, and the Holocene. Thermocline dwellers and N. dutertrei were significantly higher during the Last Deglaciation and significantly lower during the Holocene, however, less notable changes were also observed within the Last Deglaciation and Holocene intervals (see Figures 5 and 6). Less notable increases of thermocline dwellers were indicated around Mid Holocene (80-120 cm and 60-65 cm intervals) and Late Holocene (0-20 cm interval) (see Figure 5). N. dutertrei was slightly increased around Mid Holocene (80-120 cm interval) and slightly decreased around Mid Deglaciation (180-200 cm interval) (see Figure 6). Mixed layer dwellers showed a contradictive pattern compared to the thermocline dwellers and N. dutertrei with lower abundance during the Last Deglaciation and higher abundance during the Holocene (see Figure 7). During the Holocene, minor decreases are detected around Mid Holocene (90-115 cm and 60-65 cm intervals) and Late Holocene (0-20 cm interval).
The higher abundance of thermocline dwellers and the lower abundance of mixed layer dwellers indicated a shallower DOT during the Last Deglaciation compared to the Holocene (see Figures 5 and 7). Minor increases of the thermocline dwellers and minor decreases of the mixed layer dwellers around the Mid and Late Holocene Sedimentation rates between the aged depth intervals (b). also indicated DOT shoaling but to a lesser extent. Paleoproductivity was higher during the Last Deglaciation and lower during the Holocene (see Figure 6). A minor decrease of paleoproductivity was indicated around the Mid Deglaciation while its minor increase occurred around the Mid Holocene (see Figure 6).

Core ST14 (western Sumba Strait)
The pattern of the thermocline and mixed layer dwellers were similar to the core ST08 but with a less obvious shift (see Figures 5 and 7). The shift of the thermocline and mixed layer dwellers occurred earlier compared to the core ST08, which was around the Late Deglaciation (~170-180 cm interval) (see Figures 5 and 7). A minor increase of thermocline dwellers and a minor decrease of mixed layer dwellers were also observed around the Late Holocene (0-30 cm interval). The abundance of N. dutertrei was more constant with a more obvious increase observed around the Mid Deglaciation (150-170 cm interval) and a less significant increase around the Mid Holocene (40-60 cm interval) (see Figure 6).
A shallower DOT was indicated during the Last Deglaciation, but a deepening of the DOT occurred earlier compared to the southwest Sumba offshore (around the Late Deglaciation). In addition, the shoaling of the DOT around the Mid Holocene was not detected (see Figures  5 and 7). Minor DOT shoaling was also indicated around the Late Holocene, as the abundance of thermocline dwellers was gradually increased, and the abundance of mixed layer dwellers decreased gradually. Paleoproductivity was mostly constant with a more significant increase around the Mid Deglaciation and a less significant increase around the Mid Holocene (see Figure 6).

Core ST13 (central Sumba Strait)
The pattern of the thermocline dwellers and N. dutertrei was relatively constant around the Last Deglaciation-Early Holocene (90 cm-bottom interval) and gradually increased afterward (0-90 cm interval) (see Figure 5 and 6). Mixed layer dwellers were less abundant during the Last Deglaciation until the earliest interval of Holocene (140 cm-bottom interval), before increasing around the Early Holocene (80-140 cm interval), and decreased gradually since the Mid Holocene (see Figure 7).
The DOT was relatively constant during the Late Deglaciation-Early Holocene before it gradually shoaled since the Mid Holocene (see Figures 5 and 7). Paleoproductivity also indicated a similar pattern which gradually increased since the Mid Holocene (see Figure 6).

Core ST12 (eastern Sumba Strait)
Core ST12 only recorded Holocene-aged sediments, indicated by the occurrence of Gl. fimbriata even in its lowest interval. The abundance of thermocline dwellers was slightly lower around the Early Holocene (100 cm-bottom interval) compared to around Late Holocene (0-80 cm interval) while a significant decrease was observed around the Mid Holocene (80-110 cm interval) and a minor decrease was observed around the Late Holocene (0-20 cm interval) (see Figure 5). The abundance pattern of mixed layer dwellers mirrors the thermocline dwellers with a significant increase around Mid Holocene (80-110 cm interval) and a minor increase around the Late Holocene (0-20 cm interval) (see Figure 7). A significant decrease of N. dutertrei was indicated around the Mid Holocene (70-140 cm interval) while the higher abundance of N. dutertrei was observed around the Early Holocene (140 cm-bottom interval) and around the Late Holocene (0-70 cm interval) (see Figure 6).
During the Holocene, the shoaling of DOT occurred twice on the eastern Sumba Strait. The more significant DOT shoaling occurred around Mid Holocene, while the less significant DOT shoaling occurred around Late Holocene (see Figures 5 and 7). A paleoproductivity decrease was indicated around the Mid Holocene, while around the Late Holocene, it was increased (see Figure 6).

The mechanism of DOT and paleoproductivity changes
Based on the analyzed proxies, the DOT at the Sumba Island offshore (southwest Sumba offshore and Sumba Strait) got shallower during the Last Deglaciation and deeper during the Holocene, as it was opposed to the DOT at the Timor Sea, the main exit passage of ITF (Xu et  A slight decrease of paleoproductivity around Mid Deglaciation (see Figure 6) was most likely not related to the upwelling intensity, as the DOT remained constant. The change of the lifted water mass characteristics was suggested as the cause for the paleoproductivity reduction. The lifted water mass was most likely the North   The earlier DOT deepening and paleoproductivity reduction (around the Late Deglaciation) in the Sumba Strait (cores ST12, ST13, and ST14) might be related to its northward position compared to the southwest Sumba offshore (core ST08), thus the AISM reactivated earlier as the Austral summer ITCZ gradually shifted southward (Wyrwoll and Miller, 2001;Kuhnt et al., 2015). Despite the lower abundance of mixed-layer dwellers on the central Sumba Strait around the Late Deglaciation, DOT shoaling did not occur due to the relatively lower abundance of thermocline dwellers. Thermocline dwellers consist of the dissolution-resistant taxa (Ravelo et al., 1990;Martínez et al., 1999), thus their relative abundance changes are more robust as a proxy for the DOT. In southwest Sumba offshore, which is located southward, the AISM reactivation most likely occurred around the Last Deglaciation-Holocene transition. The lower interval of ST14 (170 cm-bottom interval) (west-ern Sumba Strait), which indicated lower paleoproductivity, is most likely coeval to the 180-200 cm interval of ST08 (southwest Sumba offshore). This indicated that the Early Deglaciation sediment records were only available in the southwest Sumba offshore (ST08) core.
Around the Mid Holocene, DOT shoaling accompanied by higher paleoproductivity (eutrophic condition) only occurred at southwest Sumba offshore. The shoaling of DOT and eutrophic condition was most likely caused by the stronger AIWM ( In the Sumba Strait, a change of the DOT was only indicated in its eastern part, which was DOT deepening. The deepening of DOT could be related to the intensification of surface water ITF owing to stronger AIWM (Ding et al., 2013). The intense AIWM would transfer the more saline Banda Sea waters to the southern tip of the Makassar Strait, thus displaced the freshwater plug that blocked the surface water ITF (Xu et al., 2006; Ding et al., 2013). The deepening of the DOT indicated that the effect of ITF variability could reach the Sumba Strait, even though this was only in its eastern part. The deepening of the DOT also resulted in the lower paleoproductivity (oligotrophic) condition at eastern Sumba Strait, as a thicker mix layer would inhibit the eutrophication process (Brasier, 1995) and vertical mixing was not effective This mechanism explains the eutrophic conditions inferred in the central and eastern Sumba Strait which coincided with the enhancement of the riverine input proxy (ln K/Ca) (see Figure 9). This mechanism was more effective in the central and eastern Sumba Strait due to its proximity to the surrounding islands compared to the southwest Sumba offshore and the western Sumba Strait. A similar mechanism was also indicated off southwest Java during the Late Holocene, which resulted in the higher paleoproductivity condition ( The cause of the DOT deepening in the eastern Sumba Strait was most likely identical to the DOT deepening that occurred around the Mid Holocene while the intensification of surface water ITF was related to the El Niño-like condition (Hendrizan et al., 2017).
The use of the relative abundance of planktonic foraminifera as a proxy to interpret the DOT and paleoproductivity has been proven effective in the southern Indo- Despite the effectiveness, the interpreted DOT and paleoproductivity from the relative abundance of planktonic foraminifera were basically qualitative. The use of geochemical proxies from both the foraminifera tests (δ 18 O, δ 13 C, Mg/Ca, etc.) and bulk sediments (total organic carbon, carbonate content, etc.) is suggested to produce more plausible results in future studies. In addition, radiocarbon dating data of the Sumba Strait cores (ST12, ST13, and ST14) is needed to produce more robust geochronology and simplify the proxies comparison between the studied cores.

Conclusions
(1) DOT: A contrast DOT pattern compared to the Timor Sea (main exit passage of ITF) during the Last Deglaciation-Holocene was indicated at the Sumba Island offshore based on thermocline and mixed layer dwellers' relative abundances. This concluded that the glacial-interglacial hydrographic changes of ITF did not cause the DOT changes at Sumba Island offshore, instead, they were primarily driven by the AIM variability. The role of EN-SO-like and IOD-like mechanisms to the DOT changes was only indicated around the Late Holocene. The effect of ITF changes might reach the eastern Sumba Strait, which was indicated by the shallower DOT around the Mid and Late Holocene.
(2) Paleoproductivity: N. dutertrei relative abundance indicated that the monsoon-driven coastal upwelling intensification which resulted in DOT shoaling and eutrophic condition was not always the cause for the paleoproductivity enhancement. The increase of nutrient availability in surface water due to the increase of river runoff and the changes in the lifted water mass nature were also able to enhance paleoproductivity.

Acknowledgement
This research is part of the concept of sustainable coastal environmental management for eutrophication disaster mitigation research in the Priority Research and Demand Drive Research Coremap-CTI (Number: SP DIPA-079.01.1.664156/2020). We express our gratitude to Dr. Dyah Marganingrum for the permission to participate in this research. We also thank the Research Center for Oceanography of LIPI, especially Udhi Hermawan, Ph.D. as the chief scientist of E-WIN 2016 for the permission of data usage and administrative assistance and all crews of the Baruna Jaya VIII R.V., especially Singgih Adi Wibowo for technical supports and data collecting. The Research Center for Geotechnology of LIPI and Geological Engineering Department of ITB are thanked for the laboratory facilities. We also express our grati-tude to the Indonesian Endowment Fund for Education (LPDP) (grant number: 202001110215954) for the financial assistance. Istiana, Adwi Arya, Adrianus Damanik, and Ahmad Nabil are thanked for their assistance in sample preparation and observation for core ST08, ST12, ST13, and ST14. Rizky Amalia Maulidiatsani is thanked for the assistance in proofreading.