Light environment drives the shallow to mesophotic coral community transition

Light quality is a crucial physical factor driving coral distribution along depth gradients. Currently, a 30 m depth limit, based on SCUBA regulations, separates shallow and deep mesophotic coral ecosystems (MCEs). This definition, however, fails to explicitly accommodate environmental variation. Here, we posit a novel definition for a regional or reef-to-reef outlook of MCEs based on the light vs. coral community-structure relationship. A combination of physical and ecological methods enabled us to clarify the ambiguity in relation to that issue. To characterize coral community structure with respect to the light environment, we conducted wide-scale spatial studies at five sites along shallow and MCEs of the Gulf of Eilat/Aqaba (0-100 m depth). Surveys were conducted by Tech-diving and drop-cameras, in addition to one year of light spectral measurements. We quantify two distinct coral assemblages: shallow (<40 m), and MCEs (40-100 m), exhibiting markedly different relationships with light. The depth ranges and morphology of 47 coral genera, was better explained by light than depth, mainly, due to the Photosynthetically Active Radiation (PAR) and Ultra Violet Radiation (1% at 76 m and 36 m, respectively). Branching coral species were found mainly at shallower depths i.e., down to 36 m. Among the abundant upper mesophotic specialist-corals, Leptoseris glabra, Euphyllia paradivisa and Alveopora spp., were found strictly between 36-76 m depth. The only lower mesophotic-specialist, Leptoseris fragilis, was found deeper than 80 m. We suggest that shallow coral genera are light-limited below a level of 1.25% surface PAR and that the optimal PAR for mesophotic communities is at 7.5%. This study contributes to moving MCEs ecology from a descriptive-phase into identifying key ecological and physiological processes structuring MCE coral communities. Moreover, it may serve as a model enabling the description of a coral zonation world-wide on the basis of light quality data.

Therefore, even when light at the surface is equal for two locations, its quality (i.e. 93 intensity and spectrum) may differ at the same depths at those locations due to light-dependent marine organisms has also been documented at local scales, albeit on 98 much smaller scales (Manuel et al. 2013). 99 Though PAR enhances coral growth at certain depths, light can also have negative MCEs, which may result in regional or reef-to-reef definitions, we need more studies  In this study, we sought to address this knowledge gap. We provide ecological data 132 pertaining to stony coral reefs across depth and space in the Gulf of Eilat/Aqaba 133 (GoE/A), together with light quality data. The present work emphasizes the crucial 134 importance of light regimes to the structure of coral communities in both vertical and 135 horizontal space, while offering a novel model for coral reef zonation that combines 136 physical and ecological methods.

139
Data collection 140 Spatial benthic surveys were conducted over 10 km of reef at five sites along shallow 141 reefs to MCEs (0-100 m depth). Belt-transects, 50 m in length, were recorded in bins 142 between 5-100 m (5, 10, 20, 30, 40, 50, 60, 70-80, 90-100 m) parallel to the shore at 143 each site. 2,320 photo-plots (70×50 cm) from all transects were analysed for live coral 144 cover and bathymetric substrate type (e.g. rock, gravel and dead corals) and non-145 available settlement area (sand). We used the multiple points method in the 'CoralNet' 146 web interface (Available at: http://coralnet.ucsd.edu/) to estimate percentage cover in 147 the photo-plots. Thirty photo-plots were randomly selected from each 50 m transect to 148 record coral genus abundance, for later community structure analysis.

149
Light and water temperatures were recorded monthly from August 2014 to July 2015, 150 using a profiling reflectance radiometer (PRR800, Biospherical Instruments, USA), 151 which measures 19 channels (at 300-900 nm) and the integrated PAR. The instrument 152 was deployed at midday (11:00-13:00), using the free-fall technique (Waters et al. 153 1990) to maintain a vertical position and avoid shading and reflectance from the boat.

154
The measurement data were analysed using the program PROFILER (Biospherical 155 Instruments, USA). An average depth for 1% irradiance of surface UV and PAR, and 156 0.1% PAR, and PAR attenuation coefficient (KdPAR) were calculated (from 8-19 157 different GPS locations), as described by Kirk (2011). 14 years of daily Chlorophyll-a 158 concentration data were provided by the national monitoring program (NMP) in Eilat     (Table 1S). Further variability is noted between each 10 m depth interval (i.e.  (Table 2S).  May returned the lowest AIC of any month when fitting %UV to data, and March 265 when fitting %PAR (Table 3S). We therefore use these months when evaluating

278
August returned the lowest AIC of any month when fitting %UV to data, and June 279 when fitting %PAR for cluster 2 (Table 3S)

289
The light environment of the GoE/A is modulated by an annual phytoplankton bloom.

290
Time series decomposition of Chlorophyll-a data reveals that the bloom peaks 291 typically in March (Fig. 5). Note that the seasonality displayed (Fig. 5) is only plotted coefficients (Kd(PAR)) across sites (Fig. 6a). In parallel to changes in the light 302 attenuation coefficient (Kd(PAR)), there is a dissimilarity in coral abundance at the 303 different sites along the GoE/A (Fig. 6b). Overall, the limits of genera and species structure between 30-40 m (Fig. 4). Cluster 1 comprised mainly 'shallow' genera.

361
Although our statistical analysis does not allow a clear statement with respect to 362 coral assemblages differing as depth-specialists or generalist taxa, the taxa in each 363 cluster do segregate loosely by depth (Fig. 4S).

364
The two assemblages exhibit different relationships with light (Fig. 4S) among sites throughout the year (Fig 5S). March and June were selected by AIC as the best explanation of shallow and 413 mesophotic community distributions, respectively, when considering PAR (Fig. 6).

414
Modern tools, such as remote sensing, may provide information on water quality and can be seen in the tight connection between different stony corals and the 1% PAR 481 and UV limits (Fig. 2). The changing depth distributions for these species among sites 482 appears connected to light (Fig. 6)