A comparison between the 2010 and 2016 El-Ninō induced coral bleaching in the Indonesian waters

Severe coral bleaching events are always associated with El-Ninō phenomenon which caused a rise in ocean temperature between 1-2°C and that they potentially kill the corals worldwide. There were at least four severe coral bleaching events occurred in the Indonesian waters. This study aims to compare the coral bleaching events of the 2010 and 2016 and their impact on corals in Indonesian waters. Long-term (2002-2017) remotely sensed night time sea surface temperature (SST) data acquired from Aqua MODIS Satellite were used in the analysis. Here, we calculated the mean monthly maximum (MMM)of SST as SST in normal condition in which coral can adapt to temperature; the differences between high SST in each pixel during coral bleaching events of the 2010/2016 and MMM SST, called hot spot (HS); and how long has HS occupied a certain water body, called degree of heating weeks (DHW, °C-week) and then mapped it. Results show that the MMM SST for the Indonesian waters is 29.1°C. Both bleaching events of 2010 and 2016 started and finished in the same periods of Mar-Jun and they nearly have the same pattern, but bleaching magnitude of the 2016 was stronger than 2010 with the mean SST about 0.4°C higher in May-June. The percentage of impacted areas of strong thermal stress on corals of Alert-1 plus Alert-2 status was higher in 2016 (39.4%) compared to 2010 (31.3%). Coral bleaching events in the 2010 and 2016 spread in almost all Indonesian waters and relatively occurred in the same places but with small variation in the bleaching sites that was caused by the strength/weakness of El-Ninō and upwelling phenomenon as well as the role of Indonesian through flow (ITF).

There are four significant global coral bleaching events due to global warming induced by the El Ninō which have been reported all over the world including in the Indonesian waters. However Table 3).

Source of SST data
Since coral bleaching events are caused by global ocean temperature rise that are triggered by El-Nino, monitoring the ocean temperature (SST) effectively and efficiently using remote sensing techniques through the utilization of satellite data is very important. The advantages of satellite-derived SST are vast in coverage at high resolution compared to any other form of conventional collected SST data [39]. Among the satellite derived SST products, MODIS has been providing high quality global SST for over a decade (~15 years) from a single sensor [39,40] and used for a wide variety of studies in the field of earth's climate system, weather forecasting, and oceanographic research, e.g., ocean circulation modelling and the complexity of ocean surface currents, large scale SST anomalies that indicated climate perturbations such as El-Niño events, upwelling regions, ocean biology, including coral reefs and algae blooms [39][40][41][42] and many other wide-range topics. Therefore, in this study, we used monthly SST data from 2002 to 2017 (15 years) derived by Thermal infrared (TIR) sensor of MODIS (Moderate Resolution Imaging Spectroradio-meter) at 11 μm bands of Aqua satellite with a ground resolution of 4 km by 4 km intensively. These SST data are available in the Giovanni (Geospatial Interactive Online Visualization ANd aNalysis Infrastructure) online data system web which are developed and maintained by the NASA GES DISC [43]. The SST data used in the analysis are night time SST to avoid sun glare and wide variations of SST during day time [12]. SST data used covered all Indonesian waters from latitude of 12° South and 8° North, and longitude of 92-142° East. These observed regions are also included as a part or whole region of neighboring countries such as Malaysia, Singapore, Brunei and Timor Leste.

Data analysis
The principal analysis in this study followed the standard method of NOAA which is described in details in [12]. Firstly, we collected long-term (2002-2017) night time SST data from Giovanni Web. Secondly, we calculated maximum mean monthly night SST (MMM) or SST at the normal condition which assumes that corals can adapt to seawater temperatures in these long periods (15 years) data. Third, we then calculated the differences between high night time SST due to the impact of El Niño during bleaching events of the 2010 and 2016 and MMM SST (SST in the normal condition), called hot spot (HS). Hence, HS = SST in bleaching period of 2010/2016 -MMM SST. The next step is to determine how long the HS occupied a certain water body, called degree Heating Weeks (DHW, weeks-˚C). Based on the HS and DHW values, the coral bleaching alert system, as well as the monthly maps of HS and accumulated DHW of Mar-Jun 2010/2016 for Indonesian waters can be generated for determining the impacts of coral bleaching in a particular site. Figure 1 shows the flow chart of the procedure. The performance of study on the global scale SST anomalies that indicated climate perturbations such as El-Niño events always depend on the accuracy of SST measured by the sensor which means accurate global measurements of SST are critical to understanding the past, current and future climate change [40]. Therefore, the satellite sensor must be stable to conduct long time measurement of SST, must be capable to detect the small changes of trends as small as 0.1°C within a decade, and must be validated using SST measured from ships and other platforms (Argo buoy) as ground truth [44].

Figure 1.
Flow chart of methodology and criteria in determining the status of coral bleaching [12].
Results of SST validation measured by MODIS sensors agreed well with in situ buoy SST of the Yellow Sea coastal waters, China with squared correlation coefficients (R 2 ) of 0.987, a bias of 0.06 °C, a standard deviation (STD) of 0.85 °C, and a root mean square error of 0.85 °C [45]. MODIS derived-SST achieves an excellent performance at night-time with accuracies to the order of 0.35 °C compare to in situ SST measurements derived from ARGO buoy data in the Canary Islands-Azores-Gibraltar area [46]. The SST derived from MODIS sensor in the South China Sea have biases ranging from -0.19°C to -0.34°C and STD errors ranging from 0.58 °C to 0.68 °C [47]. Study of [41] shows that the MODIS SST are comparable in accuracy to the SST derived from the Advanced Very High-Resolution Radiometer (AVHRR) sensor of Pathfinder satellite of NOAA. The AVHRR sensor data have been extensively used by NOAA for generating the global and regional scale of monthly or seasonal coral bleaching warning system and prediction maps. Thus, based on many validations, it is no doubt that sensor MODIS of Aqua satellite can also be used for this study.  Figure 3 shows the monthly map of HS distribution from March to June which also indicated that HS in 2016, especially the warning status (HS > 1 ºC, DHW ≤ 4 ºC-week), was distributed wider than in 2010. Thus, that indicated that the overall magnitude of HS 2016 was stronger than that of 2010.  In March 2010, the no thermal stress ststus (HS ≤ 0ºC) covered very wide areas at about ⅓ (31.2%) of the total study site areas (8ºN-12ºS; 91-142ºE) and spread mostly in the north part of Indonesia. This status distributed in the South China Sea, Sulawesi, Halmahera, Maluku, Seram Seas, which was influenced by South China Sea water masses in the west part and tropical Pacific Ocean (TPO) water masses in the middle and east part of Indonesian waters (Figure 3). No thermal stress status areas decreased in April (18.0%) and May (15.3%), but increased again in June (29.1%) with almost the same areas as in March 2010 (Table 2). However, the distribution of this status in June was in very different pattern compared to the one in March 2010, which spread mostly in all south part of Indonesian waters such as the south waters of Jawa, Bali, West and East Nusa Tenggara (NTB and NTT) Islands, and Arafura Seas that are under the influences of the Indian ocean water masses ( Figure  3). Low SST of the no thermal stress status in June 2010 was due to strong upwelling that pushed cold water masses from the Indian Ocean to the Arafura and Banda Seas and even farther into the Seram and Maluku Seas [31]. Upwelling is a blessing for coral bleaching events. On a local scale, small upwelling (several tens to several hundred square meters) can reduce coral bleaching events by decreasing temperatures or fluctuating environmental temperatures, so over time corals become more resistant to temperature [11]. As an example, upwelling that occurred in Tayrona National Park, Colombia in Nov-Dec 2010 has reduced the intensity of the coral bleaching events by lowering the SST from 28 °C to 21 °C between December 2010 and February 2011 [48].  (Table 2). In June, warning status spread wider in the west coast of Sumatra Island from Aceh down south to Padang that covered many islands (Nias to Mentawai Islands), Malacca and Karimata Straits, East Kalimantan (Derawan Islands) and the offshore of Cendrawasih Bay with percentage cover areas of 23.4%. The Alert-1 status (DHW 4-8 ºC-week) areas tended to rise from 0.1% in March to 1.3% in June, but almost none (~ 0 %) for Alert-2 status (DHW > 8 ºC-week).

Distribution of HS in
In   (Figure 4) [49]. However, there are little local variations of coral bleaching from one place to another. These variations may be due to the role of Indonesian through flow (ITF) that provides a sea link between the tropical Pacific Ocean (TPO) and Indian Ocean (IO) which is composed of the intricate patterns of passages and seas of varied dimensions [50]. The ITF carries warm tropical waters from the Pacific to the Indian Oceans which regulates the coupled ocean and atmosphere climate system with linked to ENSO (El-Niño and La Niña) Phenomenon and Asian monsoon, and affects regional circulation of the ocean, SST pattern and marine ecosystems [51,52].   The warning and Alert-I status in Figure 5 had higher coverage areas in the 2010 (17.2%) compare to 2016 (15.4%). On the contrary, the Alert-II status, in which corals were exposed to thermal stresses which may result in wide bleaching and potentialy kill corals of 2016 events, had wider coverage areas (24.0%) than 2010 (14.1%) (Table2). The corals severity distribution in Figure 5 shows also the differences magnitude of bleaching events between 2016 and 2010, which agrees well with the bleaching severity information collected from various sources as listed in Table 3 and Figure 6.
In the shallow water of Bunaken Island, [6] reported in their paper that almost all reef flats showed evidence of mortality of Porites, Heliopora and Goniastrea corals, representing 30 % of Bunaken reefs. Mortality was related to sea level variations, with increased aerial exposure time during a few months. In September 2015 the sea level was at its lowest in the past 12 years as observed using remotely sensed satellite altimeter data [6]. However, our results show that corals in Bunaken (site #15) were free from severe bleaching events both in 2010 and 2016, since they do not appear on the maps of Figure 5 with status of no thermal stress on corals (HS < 0˚C). Meanwhile, in the eastern of Arafura Sea, coral bleaching did not occur in the 2010 due to strong upwelling phenomenon [31] as well as in the 2016, although upwelling was weakened by strong or severe bleaching in the south of Timor Island ( Figure 5) that reached to north Australia as reported in [20].
The variation sites of bleaching events/non-bleaching events were due to the strength/weakness of ITF that was influenced by El Niño/La Niña phenomenon and, in some cases, reinforced by local airsea interactions [51,54]. For example, the TPO warm water masses that enter the South China Sea (SCS) through Luzon Strait throughflow (LS-TF) are large during El Niño, while small during La Niña. Those water masses flow southward to Karmata Strait then enter to the Jawa Sea and move eastward to reach the Makassar Strait [51]. The pattern of water masses movement in the SCS as stated by [51] is in line with both of the DHW maps of Mar-Jun 2010 (weaker condition) and Mar-Jun 2016 (stronger condition) as clearly seen in Figure 5.
The massive coral bleaching events in the Indonesian waters show a tendency that the period of bleaching is becoming shorter, i.e. four severe bleaching events reported firstly in the 1983/83 took 14-15, 12, and only 6 years to next the bleaching event in 1997/98, 2010 and 2016, respectively. Coral bleaching also tends to spread more widely and with stronger magnitudes, resulting in dramatic changes to the reefs that lead to corals extinction [8]. Modeling result on coral bleaching indicates that if corals cannot withstand the sea temperature rises of 0.2-1.0°C/decade then within 30-50 years ahead the bleaching event will occur every two years or even annually [10] while [5] stated that the bleaching phenomenon every two years have been in sight since the 1980s. The decreasing of coral reefs due to coral bleaching would have direct consequences especially to local communities which depends on part or on the whole on coral reefs' goods and environmental services (coastal protection, food security and tourism) for supporting their livelihood [6,59,60].  Table 3. Circles, respectively indicate the occurence of coral bleaching in the Indonesian waters (white), Malaysian (pink), Singaporean (cyan), and the sites that coral bleaching did not occur (blue). See Figure 3 for legend.

Conclusions
The El-Ninō induced coral bleaching in the Indonesian waters in the 2010 and 2016 have been studied using long-term (2002-2017) monthly night SST data acquired from Aqua-MODIS satellite. The MMM SST or SST in the normal condition for corals to adapt in the Indonesia waters is 29.1˚C. The differences of high SST induced by El-Ninō of 2010 and 2016 and the normal MMM SST were mapped. Based on these maps, the coral bleaching events can be recognized from start to end in the same month period of March and June (4 months period). Outside of those periods, there are no bleaching events detected. However, bleaching events of the 2016 had higher magnitude compared with the 2010 and reached its peak in May to June. The distribution of coral bleaching severity at Alert-1 and Alert-2 covered total areas of study sites of 31.3% and 39.4% in the bleaching events of the 2010 and 2016, respectively. The severity map of coral bleaching generated in this study can pinpoint the locations of bleaching events both from our field observations and also from information obtained from many sources.  Generating such kind of map by using sensor (MODIS, AVHRR) with higher resolution (1 km by 1 km) will improve the analyses. Thus, this method can be used efficiently in COREMAP-LIPI program to monitor coral bleaching events, to analyze the impacts caused on coral reef ecosystem, and to make proper diagnostic for conducting mitigation after climate-induced disturbances. Since there is a tendency that the period of coral bleaching to become shorter in the near future, it is necessary to observe the recovery rate of coral reefs as well as its associated components affected by coral bleaching.