Environmental Data Mining to Determine Coral Bleaching Severities

doi:10.21203/rs.3.rs-1840310/v2

Download PDF

Research Article

Environmental Data Mining to Determine Coral Bleaching Severities

https://doi.org/10.21203/rs.3.rs-1840310/v2

This work is licensed under a CC BY 4.0 License

Version 2

posted

You are reading this latest preprint version

Coral bleaching events could be viewed as a resultant of complex interactions involving various environmental factors. Just recently, a TEL (seawater Temperature-ENSO-Longitude) model is developed to predict the global bleaching events during the single 2016 El Niño episode. The model has a modest skill of approximately 50% predictability. In an attempt to improve coral bleaching prediction skill, here, two models are developed: an extended TEL model and a statistical model that combines multivariate and supervised classification techniques. The skills of both models are assessed in predicting bleaching events from three coral regions during several El Niño episodes. Two important findings are found. First, it is revealed that seawater salinity, ENSO, and Madden Julian Oscillation have stronger effect than seawater temperature in coral bleaching events. This has an important consequence to the fate of coral ecology when these phenomena are also subjected to change under global warming. Second, the predictability of our developed is 74% which is higher than that of the TEL model. Our model could serve as a tool for reef managers on communicating bleaching risk earlier. It could also be used for monitoring and reservation plans to mitigate the detrimental effect of various environmental stressors on corals.

Coral bleching severities

Stepwise regression

Principal Component Analysis

Decision Tree

Data Mining

Coral reefs, in combination with sea-grasses and mangroves ecosystems, provide goods and services to their coastal environments (Moberg and Folke, 1999). In combination with sea-grasses and mangroves ecosystems, coral reefs effectively serve as coastal protectors by attenuating storm, current and wave energy (Guannel et al., 2016). Numerous goods are also harvested from coral reefs. They include but not limited to: construction and marine aquaria materials (Guzmán et al., 2003; Caras and Pasternak, 2009), high-valued ornamental fish (Bruckner, 2005), and natural products (Cooper et al., 2014). These precious ecosystems, however, are experiencing serious pressures due to bleaching when corals expel their endosymbiotic algae called zooxanthellae due to environmental stresses (Sully et al., 2019).

Frequent bleaching events in recent years threatens these important coral ecosystems at various locations such as those in Coral Triangle (McManus et al., 2020), Great Barrier Reef (Curnock et al., 2019), and Caribbean (Muñiz-Castillo et al., 2019). The stresses could be triggered by rising of seawater temperature (Ainsworth et al., 2016; Palmer, 2018) caused by El Niño (Claar et al., 2018) and lowering of seawater salinity (Aguilar et al., 2019) due to excessive rainfall (Llovel et al., 2019). There are several thermal metrics for determining a temperature threshold for bleaching to occur. They are: Degree Heating Days/DHD (Lee et al.. 2020), Degree Heating Weeks/DHW (Kayanne, 2017) and Degree Heating Months/DHM (Spillman and Smith, 2021). These metrics are important in many respects include providing an early warning risk of incoming high thermal stress to any reef manager (Liu et al., 2018). Recently, McClanahan et al. (2019) develops a TEL (seawater temperature and longitude) index to predict global coral bleaching events at during a single ENSO event. They show that this index has a much higher predictability of 47% which is much higher than that of the DHW index of 9% predictability.

In this work, an extended TEL-like model and a model that uses data mining techniques are developed and tested on predicting coral bleaching severities in three coral areas, i.e. Caribbean, Great Barrier Reef and Coral Triangle. These bleachings took place during several ENSO episodes. They are: the strong El Niño of the 1987-88 and 1991-92, the very strong El Niño of the 1982-83 and 1997-98, and couple of strong La Niña of the 1988-89, 1999-00, 2000-01, 2007-08, and 2010-11 (Null 2020). Both models skills are compared to the original TEL model (McClanahan et al. 2019). We then present implications of this work on coral reservations.

2.1. Study sites

There are 3 (three) coral areas studied in developing and testing the models. They are: Caribbean, Great Barrier Reef, and Coral Triangle. The ENSO episodes during the bleaching events are: the 1983–2010 Caribbean bleaching events, the 1994–2010 Coral Triangle bleaching events, and the 1998–2010 Great Barrier Reef bleaching events.

2.2. Data acquisition

The public domain environmental data as model inputs are downloaded from different websites. The data inputs or predictors for the model are monthly data: local SST (sea surface temperature) and SSS (sea surface salinity), remote indices such as: SSN (sunspot Numbers), ENSO, PDO (Pacific Decadal Oscillation), NAO (North Atlantic Oscillation), IOD (Indian Ocean Dipole), Monsoon, and MJO (Madden Julian Oscillation). In order to capture corals acclimatization and responses to environmental stressors (Thornhill et al. 2006), lagged in time for the inputs are used. They are from the month of bleaching to the previous 12 months before the bleaching event. For instance, environmental inputs value at the month of bleaching and value at the previous 12 month has a subscript ‘0’, and ‘-12’, respectively (see Table 1). The model output or predictant is the monthly coral bleaching severity (Donner et al. 2017). They are: category ‘1’ or ‘Mild’ (1–10% bleached), category ‘2’ or ‘Moderate’ (11–50% bleached), and category ‘3’ or ‘Severe’ (> 50% bleached). The input and output data set can be found in the Supplementary section entitled Coral bleaching and Environmental data sets.

2.3. Data modeling and prediction

There are 2 (two) coral bleaching models developed in this paper. First, the TEL model of McClanahan et al. (2019) is extended by adding lagged-in-time seawater temperature and ENSO data as model inputs. These lagged-inputs are expected to represent corals acclimatization and stress responses (Thornhill et al. 2006. Then, predictor/predictant pairs are used to develop a simple multiple regression (depicted as MR in Fig. 1) model using the ‘REGRESS’ subroutine in MATLAB®.

Second, a statistical model is developed using the predictors/predictant pairs. These pairs are passed through the combination of SWMR (stepwise multiple regression), PCA (principal component analysis), and DT (decision trees) techniques. The SWMR selects the significant inputs or predictors based on their p-value of 0.05 or less using the ‘STEPWISELM’ subroutine in MATLAB®. The strength of each selected input is determined based on its standardized β coefficients, i.e. the strongest predictor appears first in Table 1. Then, a dimensional reduction is performed against these selected inputs using the ‘PCA’ subroutine in MATLAB®. In this model, three principal components that explain more than 99% of the data variance are retained. The DT algorithm is then uses PCA scores of the first and the second principal components scores to classify bleaching severity and plot the classification results in 2-dimensional map. Outputs classification and visualization are obtained using the ‘FITCTREE’, ‘RESUBPREDICT’, and ‘VIEW’ subroutines in MATLAB®.

2.4. Prediction skill assessment

The skills of both models described above in predicting coral bleaching category are measured using predictability – the coefficient of determination R², accuracy (ACC) or Proportion Correct, and Heidke Skill Score (HSS; Makridakis et al. 1998; Jolliffe and Stephenson 2012). The HSS is computed using a ‘HEIDKESS’ subroutine (Lukas 2020). Prediction with a perfect skill has the HSS value of 1 while that of no-skill prediction has the HSS of 0.

Figure 1 describes two coral bleaching models, the TEL-like and our combined model. The former has SST (Sea Surface Temperature), ENSO, and longitude that directly fed into an MR (Multiple Regression) model to predict bleaching categories. Our model has both local forcing such as seawater temperature (SST) and salinity (SSS – Sea Surface Salinity), and a couple of remote ocean-atmospheric phenomena used as model inputs or predictors. The remote factors are chosen specifically depending on the area where the bleach occurs. They are: El Niño Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), North Atlantic Oscillation (NAO), Indian Ocean Dipole (IOD), Madden Julian Oscillation (MJO), Monsoon and Sunspot numbers (SSN). These remote forcing, known to affect rainfall patterns over the studied areas, may induce sea water temperature and salinity changes within the areas where corals live (Malmgren et al. 1998; Giannini et al. 2000; Taylor et al. 2002; Rodriguez et al. 2014; Fan et al. 2019; Rensch et al. 2019; Zhuang and Duan 2019; Naidu et al. 2020; Ratnam et al. 2020). These predictors are then processed through a stepwise multiple regression (SWMR), a Principal Component Analysis (PCA), and a decision tree (DT) supervised learning in order to predict coral bleaching categories (see Methods). There are three type of bleaching severities (Donner et al. 2017). They are: category ‘1’ or ‘Mild’ (1–10% bleached), category ‘2’ or ‘Moderate’ (11–50% bleached), and category ‘3’ or ‘Severe’ (> 50% bleached).

Table 1 shows 2 (two) important results. First, significant factors, determined by their standardized coefficient β’s, responsible for bleaching events are different for each bleaching region (Fig. 2). For example, all remote factors but seawater temperature play significant role in the Caribbean coral bleaching events. Seawater temperature has a minor contribution in the Great Barrier Reef and the Coral Triangle bleaching events. It is also found that ENSO, MJO and PDO from distant past and seawater salinity near the time of bleach are strong correlates to coral bleaching. It is important to note that seawater salinity, ENSO and seawater temperature responsible for the Great Barrier Reef bleaching events have different sign at different time lag – this seems to be a trigger mechanism for a bleaching to occur.

Second, using the HSS (Heidke Skill Score) as prediction skill metric, our statistical model developed here remarkably has a much higher skill than that of the extended TEL model except for the Coral Triangle bleaching case (Table 1 and Fig. 2). Our statistical model performance for classifying coral bleaching severities are shown in Fig. 3.

Table 1

Predictors used in the extended TEL and combined models presented in Fig. 1. The models are used for predicting the Caribbean, Coral Triangle, and the Great Barrier Reef (GBR) bleaching event categories. The subscript in each predictor is the lagged in month from the time when the bleaching occurs. For instance ENSO_− 12 and ENSO₀ means the ENSO value at 12 months before the month of bleaching and the ENSO value at the month of bleaching, respectively. The model’s skill is determined using three metrics: the determination coefficient R², the forecast accuracy ACC, and the Heidke Skill Score HSS described in section 2.4.
Region/ Events	Predictors	Selected predictors	Prediction skill
			R²	ACC [%]	HSS
Caribbean: 67 Mild, 23 Moderate, and 26 Severe events	ENSO₀,…, ENSO_− 12, SST_o,…,SST_− 12, Longitude.	All selected	0.23	50.00	0.22
Caribbean: 67 Mild, 23 Moderate, and 26 Severe events	ENSO₀,…, ENSO_− 12, PDO₀,…, PDO_− 12, NAO₀,…, NAO_− 12, SST_o,…,SST_− 12, SSS_o,…,SSS_− 12, SSN_o, …,SSN_− 12.	ENSO_− 11 (β = 0.84), ENSO_− 10 (β = 0.63), PDO_− 11 (β = 0.32),NAO_− 3 (β=-0.32), NAO₀ (β = 0.21), ENSO₀ (β = 0.09), SSS₀ (β=-0.04), SSN₀ (β=-0.01)	0.39	76.72	0.58
Coral Triangle: 11 Mild, 14 Moderate, and 27 Severe events	ENSO₀,…, ENSO_− 12, SST_o,…,SST_− 12, Longitude.	All selected	0.67	78.85	0.66
	ENSO₀,…, ENSO_− 12, IOD₀,…, IOD_− 12, PDO₀,…, PDO_− 12, MJO₀,…, MJO_− 12, Monsoon₀,…, Monsoon_− 8, SST_o,…,SST_− 12, SSS_o,…,SSS_− 12, SSN_o,…,SSN_− 12.	ENSO_− 12 (β = 0.37), MJO_− 11 (β=-0.34), SST_− 1 (β=-0.15)	0.60	76.92	0.63
Great Barrier Reef: 28 Mild, 16 Moderate, and 39 Severe events	ENSO₀,…, ENSO_− 12, SST_o,…,SST_− 12, Longitude.	All selected	0.78	81.93	0.72
	ENSO₀,…, ENSO_− 12, SST_o,…,SST_− 12, SSS_o,…,SSS_− 12, SSN_o,…,SSN_− 12.	SSS_− 2 (β = 287.20), SSS_− 1 (β=-248.20), ENSO_− 12 (β=-0.5395), SST₀ (β=-0.06), SSN_− 1 (β = 0.02), SSN_− 4 (β= -0.02)	0.74	85.54	0.77

Figure 3 shows the classification map based on the DT (decision Tree) algorithm using PCA-1 and PCA-2 scores as inputs for predicting the Caribbean, the Coral Triangle and the Great Barrier Reef coral bleaching events, respectively. Some misclassifications are found in Fig. 3. The Caribbean has 27 misclassifications while GBR and Coral Triangle both have 12 misclassifications. The skill of these combined models could be improved by adding more physical quantities as inputs. They are: water surface currents (DeCarlo and Harrison 2019), water turbidity (Sully and Woesik 2019), water depth (Schmarek et al. 2018), internal waves (Wyatt et al. 2020), the Atlantic Multi-decadal Oscillation (AMO; Mendez and Magaña 2010), and the Atlantic and Pacific ITCZ (Inter Tropical Convergence Zone; Martinez et al. 2019). In addition, biological aspects are also incorporated as model inputs. They are: the composition of corals and their occupants such as symbionts and bacterias (Thornhill et al. 2006; Baker et al. 2013; Smith et al. 2017; Yang et al. 2017; Cunning et al. 2018; Carballo et al. 2019; Rouze et al. 2019).

In spite of its poor predictive capability in terms of the seawater temperature DHW index (McClanahan et al. 2019), the index is still considered as the prime factor for coral bleaching (Palmer 2018; Skirving et al. 2019; Smith and Spillman 2019; Sully et al. 2019). The metric has also been used in coral monitoring (Heron et al. 2016) and coral bleaching early warning (Liu et al. 2018). Beside seawater temperature, there are other stressors responsible for bleaching such as: light (Bellworthy and Fine 2017), seawater salinity (Aguilar et al. 2019) and ocean current (DeCarlo and Harrison 2019). In this work, seawater salinity, ENSO, MJO, PDO, and NAO (see Table 1 and Fig. 2) are more important than seawater temperature on determining coral bleaching severities. It is important to note that these factors are shown to be affected by global warming (Hahn et al. 2018; Llovel et al. 2019; Bui and Maloney 2019; Yan et al. 2020; Li et al. 2020).

Predicting a global scale coral bleaching event should be conducted cautiously. Each corals area has a unique response to environmental stressors and trigger clues due to their own biological and physical characteristics as well as their specific location. The models developed here could serve as a benchmark model against any coral bleaching prediction models. Owing to its lagged-inputs up to 11 months earlier, the model could also assist reef managers by providing both early warning of an upcoming bleaching. This enables a decision maker to design monitoring and reservation plans ahead of times.

Acknowledgement:

We thank the following institutions and individuals for providing us with the online data and graphic tool: National Oceanic and Atmospheric Administration (SST, SSS, ENSO, IOD, PDO, and NAO), Koninklijk Nederlands Meteorologisch Instituut (SSN and MJO), Prof. Jianping Li (Monsoon), and Prof. S. Donner (coral bleaching data set), and Mr. Derek Watkins (Geo Coordinate Point Plotters @dwtkns.com). We also express our gratitude to Mathworks for providing us with the MATLAB® subroutines for this modeling work.

Aguilar C, Raina J, Fôret S et al (2019) Transcriptomic analysis reveals protein homeostasis breakdown in the coral Acropora millepora during hypo-saline stress. BMC Genomics 20:148 https://doi.org/10.1186/s12864-019-5527-2.
Ainsworth T D, Heron S F, Ortiz J C, Mumby P J, Grech A, Ogawa D, Eakin M and Leggat W (2016) Climate change disables coral bleaching protection on the Great Barrier Reef. Science 352 (6283): 338-342. https://doi.org/10.1126/science.aac7125.
Baker AC, McClanahan TR, Starger CJ, Boonstra RK (2013) Long-term monitoring of algal symbiont communities in corals reveals stability is taxon dependent and driven by site-specific thermal regime. Mar Ecol Prog Ser 479:85-97 https://doi.org/10.3354/meps10102.
Bellworthy J, Fine M (2017) Beyond peak summer temperatures, branching corals in the Gulf of Aqaba are resilient to thermal stress but sensitive to high light. Coral Reefs 36:1071–1082 https://doi.org/10.1007/s00338-017-1598-1.
Bruckner A W 2005 The importance of the marine ornamental reef fish trade in the wider Caribbean Rev. Biol. Trop. 53: 127-137. https://www.scielo.sa.cr/scielo.php? script=sci_arttext& pid=S0034-77442005000300018.
Bui HX, Maloney ED (2019) Mechanisms for Global Warming Impacts on Madden–Julian Oscillation Precipitation Amplitude. J Clim 32(20):6961–6975 https://doi.org/10.1175/JCLI-D-19-0051.1.
Caras T and Pasternak Z (2009) Long-term environmental impact of coral mining at the Wakatobi marine park, Indonesia. Ocean Coast Manage 52 10 539-544. https://doi.org/10.1016/j.ocecoaman.2009.08.006.
Carballo BR, Denis V, Huang YY et al (2019) Temporal variation and photochemical efficiency of species in Symbiodinaceae associated with coral Leptoria phrygia (Scleractinia; Merulinidae) exposed to contrasting temperature regimes. PLoS ONE 14(6):e0218801 https://doi.org/10.1371/journal.pone.0218801.
Claar D C, Szostek L, McDevitt-Irwin J M, Schanze J J and Baum J K (2018) Global patterns and impacts of El Niño events on coral reefs: A meta-analysis PloS ONE 13 2 e0190957. https://doi.org/10.1371/journal.pone.0190957.
Cooper E L, Hirabayashi K, Strychar K B and Sammarco P W 2014 Corals and their potential applications to integrative medicine. Evid Based Complement Alternat Med 184959 https://doi.org/10.1155/2014/184959.
Cunning R, Silverstein RN, Baker AC (2018) Symbiont shuffling linked to differential photochemical dynamics of Symbiodinium in three Caribbean reef corals. Coral Reefs 37:145–152 https://doi.org/10.1007/s00338-017-1640-3.
Curnock, M.I., Marshall, N.A., Thiault, L. et al. Shifts in tourists’ sentiments and climate risk perceptions following mass coral bleaching of the Great Barrier Reef. Nat. Clim. Chang. 9: 535–541 (2019). https://doi.org/10.1038/s41558-019-0504-y.
DeCarlo TM, Harrison HB (2019) An enigmatic decoupling between heat stress and coral bleaching on the Great Barrier Reef. PeerJ 7:e7473 https://doi.org/0.7717/peerj.7473.
Donner SD, Rickbeil GJM, Heron SF (2017) A new, high-resolution global mass coral bleaching database. PLoS ONE 12(4):e0175490 https://doi.org/10.1371/journal.pone.0175490.
Fan L, Xu J, Han L (2019) Impacts of Onset Time of El Niño Events on Summer Rainfall over Southeastern Australia during 1980–2017. Atmosphere 10(3):139 https://doi.org/10.3390/atmos10030139
Giannini A, Kushnir Y, Cane MA (2000) Interannual variability of Caribbean rainfall, ENSO, and the Atlantic Ocean. J Clim 13(2):297–311 https://doi.org/10.1175/1520-0442
Guannel G, Arkema K, Ruggiero P and Verutes G (2016) The Power of Three: Coral Reefs, Seagrasses and Mangroves Protect Coastal Regions and Increase Their Resilience PLoS ONE 11 7 e0158094 https://doi.org/10.1371/journal.pone.0158094
Guzmán H, Guevara C and Castillo A 2003 Natural Disturbances and Mining of Panamanian Coral Reefs by Indigenous People Conserv. Biol. 17 5 1396-1401 http://dx.doi.org/10.1046/j.1523-1739.2003.02308.x
Hahn LL, Ummenhofer CC, Kwon YO (2018) North Atlantic Natural Variability Modulates Emergence of Widespread Greenland Melt in a Warming Climate. Geophys Res Lett 45(17):9171-9178 https://doi.org/10.1029/2018GL079682.
Heron SF, Johnston L, Liu G et al (2016) Validation of Reef-Scale Thermal Stress Satellite Products for Coral Bleaching Monitoring. Remote Sens 8(1):59 https://doi.org/10.3390/rs8010059.
Jolliffe IT, Stephenson DB (2012) Forecast Verification - A Practitioner's Guide in Atmospheric Science 2nd. John Wiley & Sons
Kayanne H (2017) Validation of degree heating weeks as a coral bleaching index in the northwestern Pacific Coral Reefs 36: 63–70. https://doi.org/10.1007/s00338-016-1524-y
Lee, IH., Fan, TY., Fu, KH. et al. (2020) Temporal variation in daily temperature minima in coral reefs of Nanwan Bay, Southern Taiwan. Sci Rep 10, 8656. https://doi.org/10.1038/s41598-020-65194-8
Li S, Wu L, Yang Y et al (2020) The Pacific Decadal Oscillation less predictable under greenhouse warming. Nat Clim Chang 10:30–34 https://doi.org/10.1038/s41558-019-0663-x.
Liu G, Eakin CM, Chen M et al (2018) Predicting Heat Stress to Inform Reef Management: NOAA Coral Reef Watch’s 4-Month Coral Bleaching Outlook. Front Mar Sci 5:57 https://doi.org/10.3389/fmars.2018.00057.
Llovel W, Purkey S, Meyssignac B et al (2019) Global ocean freshening, ocean mass increase and global mean sea level rise over 2005–2015. Sci Rep 9:17717 https://doi.org/10.1038/s41598-019-54239-2.
Lukas P. Calculate HeidkeSS (https://www.mathworks.com/matlabcentral/fileexchange/55156-heidkess-ctable), MATLAB Central File Exchange. Retrieved January 9, 2020.
Makridakis SG, Wheelwright SC, Hyndman RJ (1998) Forecasting - Methods and Applications 3rd Edition. John Wiley & Sons
Malmgren B, Winter A, Chen D (1998) El Niño–Southern Oscillation and North Atlantic Oscillation control of the Puerto Rico climate. J Clim 11(10): 2713–2717 https://doi.org/10.1175/1520-0442
Martinez C, Goddard L, Kushnir Y (2019) Seasonal unimodal and bimodal precipitation pattern are controlled by the Pacific and Atlantic ITCZ and the North Atlantic Subtropical High. Clim Dyn 53:825–846 https://doi.org/10.1007/s00382-019-04616-4.
McClanahan TR, Darling ES, Maina JM et al (2019) Temperature patterns and mechanisms influencing coral bleaching during the 2016 El Niño. Nat Clim Chang 9:845–851 https://doi.org/10.1038/s41558-019-0576-8
McManus LC, Vasconcelos VV, Levin SA, Thompson DM, Kleypas JA, Castruccio FS, Curchitser EN, Watson JR (2020) Extreme temperature events will drive coral decline in the Coral Triangle Glob Change Biol 26 (4): 2120-2133 https://doi.org/10.1111/gcb.14972
Méndez M, Magaña V (2010) Regional Aspects of Prolonged Meteorological Droughts over Mexico and Central America. J Clim 23(5):1175-1188 https://doi.org/10.1175/2009JCLI3080.1.
Moberg FF, Folke C 1999 Ecological goods and services of coral reef ecosystems Ecol. Econ. 29 215–233 http://dx.doi.org/10.1016/S0921-8009(99)00009-9
Muñiz-Castillo, A.I., Rivera-Sosa, A., Chollett, I. et al. Three decades of heat stress exposure in Caribbean coral reefs: a new regional delineation to enhance conservation. Sci Rep 9, 11013 (2019). https://doi.org/10.1038/s41598-019-47307-0
Naidu PD, Ganeshram R, Bollasina MA et al (2020) Coherent response of the Indian Monsoon Rainfall to Atlantic Multi-decadal Variability over the last 2000 years. Sci Rep 10:1302 https://doi.org/10.1038/s41598-020-58265-3.
Null J. El Niño and La Niña Years and Intensities. Available online: https://ggweather.com/enso/oni.htm (accessed on 20 Jan 2020).
Palmer CV (2018) Warmer Water Affects Immunity of a Tolerant Reef Coral. Front Mar Sci 5: 253 https://doi.org/10.3389/fmars.2018.00253.
Ratnam JV, Dijkstra HA, Behera SK (2020) A machine learning based prediction system for the Indian Ocean Dipole. Sci Rep 10:284 https://doi.org/10.1038/s41598-019-57162-8.
Rensch P, Arblaster J, Gallant AJE et al (2019) Mechanisms causing east Australian spring rainfall differences between three strong El Niño events. Clim Dyn 53:3641 https://doi.org/10.1007/s00382-019-04732-1
Rodriguez RA, Grove CA, Zinke J, Pandolfi, J M, Zhao JX (2014) Coral Luminescence Identifies the Pacific Decadal Oscillation as a Primary Driver of River Runoff Variability Impacting the Southern Great Barrier Reef. PLoS ONE 9(1):e84305 https://doi.org/10.1371/journal.pone.0084305
Rouzé H, Lecellier G, Pochon X et al (2019) Unique quantitative Symbiodiniaceae signature of coral colonies revealed through spatio-temporal survey in Moorea. Sci Rep 9:7921 https://doi.org/10.1038/s41598-019-44017-5.
Schoepf V, Carrion SA, Pfeifer SM et al (2019) Stress-resistant corals may not acclimatize to ocean warming but maintain heat tolerance under cooler temperatures. Nat Commun 10:4031 https://doi.org/:10.1038/s41467-019-12065-0.
Schramek T, Merrifield M, Terrill E (2018 Depth Dependent Thermal Stress Around Corals in the Tropical Pacific Ocean. Geophys Res Lett 45:18 https://doi.org/10.1029/2018GL078782.
Skirving WJ, Heron SF, Marsh BL et al (2019) The relentless march of mass coral bleaching: a global perspective of changing heat stress. Coral Reefs 38:547–557 https://doi.org/10.1007/s00338-019-01799-4.
Smith EG, Vaughan GO, Ketchum RN et al (2017) Symbiont community stability through severe coral bleaching in a thermally extreme lagoon. Sci Rep 7:2428 https://doi.org/10.1038/s41598-017-01569-8.
Smith G, Spillman C (2019) New high-resolution sea surface temperature forecasts for coral reef management on the Great Barrier Reef. Coral Reefs 38:1039–1056 https://doi.org/10.1007/s00338-019-01829-1
Spillman, CM, Smith GA (2021) A New Operational Seasonal Thermal Stress Prediction Tool for Coral Reefs Around Australia Front. Mar. Sci. (8) 687833 https://doi.org/10.3389/fmars.2021.687833
Sully S, Burkepile DE, Donovan MK et al (2019) A global analysis of coral bleaching over the past two decades. Nat Commun 10:1264 https://doi.org/10.1038/s41467-019-09238-2.
Sully S, Woesik VR (2019) Turbid reefs moderate coral bleaching under climate-related temperature stress. Glob Change Biol 00:1– 7 https://doi.org/10.1111/gcb.14948.
Taylor MA, Enfield DB, Chen AA (2002) Influence of the tropical Atlantic versus the tropical Pacific on Caribbean rainfall. J Geophys Res 107(C9):3127 https://doi.org/10.1029/2001JC001097
Thornhill DJ, LaJeunesse TC, Kemp DW et al (2006) Multi-year, seasonal genotypic surveys of coral-algal symbioses reveal prevalent stability or post-bleaching reversion. Mar Biol 148:711–722 https://doi.org/10.1007/s00227-005-0114-2.
Wyatt ASJ, Leichter JJ, Toth LT et al (2020) Heat accumulation on coral reefs mitigated by internal waves. Nat Geosci 13:28–34 https://doi.org/10.1038/s41561-019-0486-4.
Yan Z, Wu B, Li T et al (2020) Global warming: Eastward shift and extension of ENSO-induced tropical precipitation anomalies under global warming. Sci Adv 6(2):eaax4177 https://doi.org/10.1126/sciadv.aax4177.
Yang SH, Tseng CH, Huang CR et al (2017) Long-term survey is necessary to reveal various shifts of microbial composition in corals. Front Microbiol 8:1094 https://doi.org/10.3389/fmicb.2017.01094
Zhuang M, Duan A (2019) Revisiting the Cross-Equatorial Flows and Asian Summer Monsoon Precipitation Associated with the Maritime Continent. J Clim 32:6803–6821 https://doi.org/10.1175/JCLI-D-18-0749.1

No competing interests reported.

SupplementaryBleachingEnvironmentaldataset.xlsx

Download PDF

Version 2

posted

You are reading this latest preprint version

Environmental Data Mining to Determine Coral Bleaching Severities

Status:

Version 2

Abstract

Figures

Introduction

Methods

2.1. Study sites

2.2. Data acquisition

2.3. Data modeling and prediction

2.4. Prediction skill assessment

Result and Discussion

Conclusion

Declarations

Acknowledgement:

References

Additional Declarations

Supplementary Files

Status:

Version 2