Climate change impacts on mesophotic regions of the Great Barrier Reef

2. Methods

The S2P3-R v2.0 downscaling approach (48) enables the study of bottom temperature climate projections. This study simulates waters between 0-50 m depth and considers areas between 30-50 m as the upper mesophotic reef. Surface temperature refers to the upper most layer of the water column and the bottom temperature refers to the layer just above the seabed (Figure 1). The model is run with 2 m vertical resolution and a 0.1 m horizontal resolution (48). Stratified locations are hypothesised to provide thermal relief at the seabed when warming conditions are occurring at the surface. We created a metric to locate areas where stratification is allowing thermal isolation of bottom waters during austral summer based on a location having a positive surface temperature anomaly above a negative bottom temperature anomaly. For example, if a grid point has a bottom temperature anomaly of -0.50°C and a surface temperature anomaly of 1°C, this would constitute an area of stratification or thermal protection. On the contrary, if a bottom temperature anomaly is 1°C and the surface temperature anomaly is 1°C, this would be an unstratified or non-thermally protected area. This method is described in the equation (1) where SST represents sea surface temperature, SBT represents sea bottom temperature, and t represents time.

thermal_protectiont=((SSTt-SSTclimatology)>0) AND ((SBTt-SBTclimatology)<0)

                  (1)

Throughout the remainder of this manuscript the term thermal protection is used to refer to this state (1). By considering only periods with negative bottom temperature anomalies we avoid ambiguity about whether any thermal stress was experienced in response to surface warming. While thermal protection may be experienced in situations not identified by this metric, we can be confident that when the number of events counted by this metric falls, potential thermal refugia are being lost, and when it falls to zero, climatological values are no longer being experienced. Model anomalies are calculated using an austral summer climatology (1980-1999). Areas of thermal protection are identified under four future climate emission scenarios using five climate models. 

2.1 Downscaling

This study uniquely involves the analysis of bottom temperature output, i.e. the temperature at the seabed, derived from the semi-dynamic S2P3-R v2.0 downscaling. CMIP6 models (49); MRI-ESM2-0 (50), EC-Earth3-Veg (51), UKESM1-0-LL (52), CNRM-ESM2-1 (53), and IPSL-CM6A-LR (54), were downscaled under climate-change scenarios; SSP1-1.9, SSP1-2.6, SSP3-7.0 and SSP5-8.5 (55). To do this the S2P3-R v2.0 model was forced with atmospheric conditions from each model and scenario as described in McWhorter et al., (56). The spatial variability between climate models was further described in McWhorter et al (57). Downscaled surface and bottom temperature outputs from a S2P3-R v2.0 simulation forced with the ERA5 atmospheric reanalysis product (58) were previously compared to Australia’s Integrated Marine Observing System (IMOS) mooring system observational bottom temperature data for validation in Halloran et al (48). Simulated bottom water temperatures tend to follow a 1:1 relationship with observed temperatures, but south of the Cape York Peninsular tend to display a cold bias of approximately one degree (48). Downscaled SSTs compared to satellite SSTs on the GBR contained a positive bias in the north and a negative bias in the south potentially due to a lack of simulated lateral advection (48). The S2P3-R v2.0 downscaling captures much of the interannual variability in SSTs with a temperature bias of < 0.5 K (48).

While recent progress in the decarbonisation of global energy systems means that SSP5-8.5 is a highly unlikely scenario, it is the only scenario that has persisted, largely unchanged, across CMIP versions and is therefore valuable as a comparator. The strong signal it provides is valuable in determining impacts at given temperature thresholds and understanding mechanisms.

2.2 Summer metrics applied to surface and bottom temperature outputs  

Typically used bleaching metrics such as Degree Heating Weeks (DHW) (59) and the number of severe bleaching events/decade (56) could not be applied in this study because the thermal stress anomalies at which corals undergo bleaching at deeper depths (> 15 m) is largely unknown (60). 

The bottom and surface temperature anomaly data were used to locate areas that contain a positive surface temperature anomaly on top of a negative bottom temperature anomaly, i.e., locations where surface warming, and therefore increased buoyancy was potentially insulating bottom waters from summer heat. Since bleaching on the GBR typically occurs during austral summer months (61), surface and bottom temperature anomalies were calculated during December, January, February, and March (i.e., July 31, 1999 – August 1, 2000) (61). The anomalies were calculated in relation to the average summer conditions from 1980-1999. The areas of thermal protection are spread across the summer months and models (Supplementary Figure 1, Supplementary Figure 2). The Great Barrier Reef Marine Park Authority (GBRMPA)(62) boundary was used to mask the values within the GBRMPA boundary for consistency.  

 2.3 Validation 

The ERA5 atmospheric reanalysis product (58) was downscaled using S2P3-R v2.0 to relate the observational data to the climate model data. Downscaled ERA5 outputs are used to explore the controls on thermal protection and to ground-truth the results (57). Analysis of the ERA5 driven simulation was identical to that of the CMIP6 driven simulations. The climatology period was set as 1980-1999 (inclusive) and the years chosen for the comparison between the climate models and observations were 2000-2019 (inclusive) (Figure 2a-2c). The areas of thermal protection or stratification were calculated as a percentage of cells for each, ERA5 and CMIP6 outputs (Figure 2a-2b). ERA5 and CMIP6 outputs were then added together to show the areas of highest agreement (Figure 2c). 

 2.4 Downscaled ERA5 based wind and tidal energy flux calculations  

Wind and tidal energy impact mixing (63) and therefore the seawater temperatures experienced by GBR corals (57). This study compares wind and tidal energy over areas of thermal protection and non-thermal protection to elucidate the primary controls on stratification. Wind and tidal energy outputs were extracted from the S2P3-R v2.0 downscaled ERA5 simulation. The energy flux calculations, including the statistical methods used for comparison, are described in McWhorter et al (57). Additive mixed effect models were used to explore differences in wind and tidal energy between the thermal protection locations and the non-thermal protection  locations using the ‘bam’ function (64) in R version 4.1.1 (65) where longitude and latitude were included as a smooth function to account for the spatial correlation of the data. Pairwise comparisons were determined using the ‘pairs’ function (66) in R version 4.1.1 (65).