The spectral and spatial distribution of light pollution in the waters of the northern Gulf of Aqaba (Eilat)

The urbanization of the shores of the Gulf of Aqaba has exposed the marine environment there, including unique fringing coral reefs, to strong anthropogenic light sources. Here we present the first in situ measurements of artificial nighttime light under water in such an ecosystem, with irradiance measured in 12 wavelength bands, at 19 measurement stations spread over 44 square km, and at 30 depths down to 30-m depth. At 1-m depth, we find downwelling irradiance values that vary from 4.6 × 10−4 μW cm−2 nm−1 500 m from the city to 1 × 10−6 μW cm−2 nm−1 in the center of the gulf (9.5 km from the city) in the yellow channel (589-nm wavelength) and from 1.3 × 10−4 μW cm−2 nm−1 to 4.3 × 10−5 μW cm−2 nm−1 in the blue channel (443-nm wavelength). Down to 10-m depth, we find downwelling irradiance values that vary from 1 × 10−6 μW cm−2 nm−1 to 4.6 × 10−4 μW cm−2 nm−1 in the yellow channel and from 2.6 × 10−5 μW cm−2 nm−1 to 1.3 × 10−4 μW cm−2 nm−1 in the blue channel, and we even detected a signal at 30-m depth. This irradiance could influence such biological processes as the tuning of circadian clocks, the synchronization of coral spawning, recruitment and competition, vertical migration of demersal plankton, feeding patterns, and prey/predator visual interactions.

Scientific RepoRts | 7:42329 | DOI: 10.1038/srep42329 behavioral patterns is thought to be related to the overlap between the emission spectra of artificial light sources and the absorption spectra of light sensing cryptocromes. (Compare Fig. 1 to http://www.ledgrowlightshq.co.uk/ chlorophyll-plant-pigments/). For example, such pigments have been shown to be involved in temporal synchronization processes in corals 10 . It has been suggested that the effects of light pollution on the reproductive physiology, migration, and foraging of marine species could possibly lead to changes in biodiversity 15 and in the composition of epifaunal communities 8 . Though some ecologists previously acknowledged the potential of artificial nighttime light to disrupt ecological systems in general 2 and coastal and marine environments in particular [16][17][18][19] , the phenomenon has only recently become widely recognized as an environmental issue 20 , and in situ measurements of light pollution and its biological and ecological consequences are lacking.
The natural cycle of light and darkness in the marine environment includes the wax and wane of moonlight, as well as the diel and seasonal patterns of sunlight. The subtle spectral changes in sunlight that take place prior to sunset have recently attracted particular interest as a result of their role in determining the hour of coral spawning 21 . Moonlight is of particular importance, since some marine animals display behavior patterns that are correlated with the synodic monthly phases of the moon 22 . Marine zooplankton have been shown to utilize moonlight for navigation 19,23 , for migration 3 , and as a trigger for biological process such as reproduction 11,21,24 , predator-prey visual interaction 25 , and photosynthesis 26 . Therefore, any perturbation of the natural illumination patterns to which aquatic organisms have adapted in the course of evolution can disrupt their fine-tuned life cycles, behavioral patterns, and physiological mechanisms 2,10,27 . Anthropogenic light sources constitute a particularly significant perturbation to nighttime underwater illumination, since they operate primarily at night (rather than during the day), and since their light intensity is comparable to that of moonlight (as opposed to that of sunlight).
Particularly vulnerable are light sensitive processes and behavioral patterns of the gulf 's marine organisms, such as the lunar timing of coral spawning, as described by Atoda 28 and as confirmed in several subsequent studies [29][30][31][32][33] , the diel expansion and contraction of tentacles 32 , the diel vertical migration of zooplankton 34 , and the feeding of diurnal corallivorous fishes 35 . Anthropogenic disturbances of the natural light patterns to which organisms have evolved to be attuned to may accelerate the decline of coral reefs that are already stressed by ocean warming, acidification, and eutrophication [36][37][38] .
The contribution of corals to the primary structural composition of the reef and to the ecology of the reef via their providing an environment for breeding and protection of various organisms, along with the economic value of the corals, has been investigated over the course of a few decades 39,40 . However, the extent and potential impacts of light pollution on corals have not been investigated. Functional processes that are affected by environmental light conditions, such as the spawning phase of coral reproduction 10,41 , may be disturbed as a result of changes in the levels of light reaching the seabed, which in turn may lead to a lack of synchronization in reproductive processes. In addition, changes in the intensity and spectral quality of artificial light may affect planulae settlement and recruitment patterns of corals in the vertical zonation region of the seabed 16,42 , as well as the distribution of coral colony morphology over the area of the reef 43 .
Given that corals are highly photosensitive, with a photoreception sensitivity threshold of blue light of ~1.2 × 10 15 photons m −2 s −1 24 , and given previous studies showing that reef structure can be strongly influenced by illumination 44 , there is a definite potential for artificial nighttime lighting to have harmful effects on reef functions and reef health 45 .
Clearly, an understanding of the impact of anthropogenic light sources is crucial for the very survival, sustainable use, management, protection, and bioremediation of important tropical ecosystems, such as coral reefs. As a primary step in this direction, the light field of these sources needs to be mapped in space and time. Previous efforts in this direction have included modeling 46 , remote sensing 6,47 , and mapping with geographic information systems (GIS) 48 . However, to the best of our knowledge, the full intensity and spectral composition of nighttime light over the surface of the water and its bathymetric distribution in the water column in a coastal ecosystem have not been previously mapped. The spectral composition of the light is especially important, since different organisms sense light differently, including some organisms that have the ability to sense wavelengths outside of the visible spectrum 2,17 .
Here we report for the first time the spatial and spectral distribution of nighttime light in the waters of the Gulf of Aqaba, a complex shoreline containing sectors of industry, commerce, and tourism, and surrounding one of the northernmost flourishing coral reefs on Earth which is protected and managed by the National Parks Authority of Israel.

The study area
The Gulf of Aqaba (Eilat) is an arm of the Red Sea, surrounded by two cities, Eilat, Israel (34°95′ /29°55′ ), and Aqaba, Jordan (35°00′ /29°50′ ), each with a population of tens of thousands of residents (~60,000 and ~160,000, respectively). In addition, these cities contain industrial complexes along the shores and commercial and military ports, as well as recreational marinas and a thriving tourist industry. As the northern reef destination closest to Europe, the coral reef is the main tourist attraction, with an extremely high number of divers per unit area of reef 49 .
The Gulf of Aqaba is located in a (semi-arid) desert area with strong solar insolation, clear skies throughout most of the year (only ~30 cloudy days yr −1 , less than 30 mm rain yr −1 ; http://www.iui-eilat.ac.il/Research/ NMPMeteoData.aspx), very minimal river runoff, low water turbidity, low re-suspension of sediments 50,51 , low nutrient concentration 52 , and low plankton biomass 50,52 . These natural conditions allow natural light, as well as different forms of artificial light, to reach and penetrate the water body to considerable depths, with 1% of the subsurface irradiance reaching depths exceeding 110 m 53 . Correspondingly, the vertical attenuation coefficient for underwater downwelling irradiance of photosynthetically active radiation (K d PAR) is low, ranging between 0.04 and 0.065 m −1 50,54,55 . As such, the water in the Gulf of Aqaba may be categorized as open ocean, oligotrophic, case I waters 50,56 . We also note that the nature of the clear sky in this gulf area may reduce the extent of sky glow 9 , though scattering of artificial light by the atmosphere will still occur to some extent.
Due to the uniquely steep shoreline, the fringing reefs in the gulf skirt the shoreline at the unusually close range of a few meters 57 . Thus, they are particularly exposed to the impact of artificial light sources from the densely populated surrounding urban conglomerations. This is in addition to the combination of tourism, pollution, intensive diving, and shoreline modification, all of which have contributed to the degradation of the gulf 's reefs and the marine ecosystems that are supported by the reefs, as has been observed over the past few decades 57,58 .

Materials and Methods
The underwater downwelling irradiance (Ed) in the Gulf of Aqaba was measured on the night of September 9 th , 2014 at 23:00 local time and on the night of August 12 th , 2015 at 22:00 local time (GMT + 3.0), using a SeaWiFS-compliant, high resolution, profiling reflectance radiometer (PRR-800; Biospherical Instruments Inc., San Diego) with 19 spectral channels in the 300-900-nm wavelength range. Irradiance in each of the 19 channels, as well as the total irradiance of photosynthetically active radiation (PAR), was measured. The PRR-800 was deployed at nighttime on moonless nights, from a boat, using the free fall technique 59 in order to avoid shade or reflectance from the boat and in order to keep the light sensor in a vertical posture.
We chose the PRR-800 and its downwelling irradiance mode, because of the ease of deployment, relatively rapid measurement rate for measurements in series at different depths, stability, and compatibility, as well as its highly sensitive sensor. We note that while downwelling irradiance might be the most appropriate variable for estimating the effect of light pollution on corals that live on the sea bed, it is not the most comprehensive measurement for estimating the effect of light pollution on species that detect light incident from all angles (e.g., zooplankton). Nevertheless, since our emphasis in conducting these first measurements of their kind is in evaluating the penetration depth of the light, the downwelling irradiance mode is the most useful for this purpose. Furthermore, since downwelling irradiance is an integral quantity, it is more likely to exhibit stable numbers above the measurement threshold (see more on the measurement threshold in this section) than radiance measurements would.
Measurements taken with a value of pitch or roll greater than 10 degrees were removed from the data analysis, in a similar fashion to Wang and Zhao 60 . The instrument was lowered with a velocity of ~0.7 m s -1 , and irradiance was recorded with a sampling frequency of 5 Hz. Sampling points were distributed throughout the Israeli part of the gulf (Fig. 2). The spatial data were analyzed and presented using the ArcGIS Version. 10.2.1 (Esri Inc.) platform. Spatial interpolation of the data in the horizontal direction was conducted using the standard inverse distance weighting (IDW) method 61 within the GIS program. We note that the specifications of the PRR-800 are such that noise equivalent irradiance is defined to be 1 × 10 −6 μ W cm −2 nm −1 or ~1 × 10 −6 μ W cm −2 nm −1 in a given wavelength channel. This irradiance measurement threshold limited our ability to measure accurate values at greater depths than we present here. Likewise, we found that even at the more highly illuminated stations, for longer wavelength light, we reached the measurement threshold within 1-2 meters below the water surface. Therefore, the red part of the spectrum is not analyzed in the current study despite its potential biological significance. Note that to the best of our knowledge, conducting measurements of light with such a low irradiance remains a technological challenge. We are unaware of any published study in which light in the red part of the spectrum with such low values of irradiance has been successfully recorded.

Results and Discussion
The downwelling irradiance (Ed) measurements in the Gulf of Aqaba are shown in Figs 3-5. In Fig. 3, the horizontal variation in downwelling irradiance is shown at three depths in the water column (1 m, 5 m, and 10 m, Scientific RepoRts | 7:42329 | DOI: 10.1038/srep42329 respectively) and in two wavelength channels [589 nm (yellow light) and 443 nm (blue light), respectively]. In Fig. 4, a vertical cross sectional map (depth versus distance from the shore) of downwelling irradiance in the same two wavelength channels is shown, encompassing measurements from stations i1, i5, i8, i12, and i15 (refer to Fig. 2 for station locations). From Figs 3 and 4, one can see that artificial light from the city of Eilat was detected from the nearest adjacent waters (e.g., station i1 in Fig. 2; 500 m from the city) out to distant points in the center of the gulf (e.g., station i15 in Fig. 2; 9.5 km from the city). In addition to light from the city of Eilat, Eilat's main port, located less than 300 m from station i4 in Fig. 2, the oil jetty terminal's offshore pier, which is also located close to station i4, and light sources from the city of Aqaba also contribute to the field of unnatural light in the gulf. From Fig. 3 and 4, there is a clear horizontal gradient of irradiance, with horizontal differences of more than 1.5 orders of magnitude in the yellow channel and 0.8 orders of magnitude in the blue channel down to 10-m depth. For example, at station i4 at a depth of 1 m under the water's surface, the measured irradiance in the 589-nm wavelength channel is 4.6 × 10 −4 μ W cm −2 nm −1 (the maximum value that was measured), while at the most remote station (station i15; 6.3 km from station i1), at the same depth of 1 m, the measured irradiance in the 589-nm wavelength channel is at the measurement threshold of 1 × 10 −6 μ W cm −2 nm −1 . However, due to the multiple light sources contributing to the irradiance at each station, the horizontal gradient is not as steep as would be expected from a one over distance squared dependence from a single localized light source. On the contrary, the irradiance measured at the more remote stations is higher than would be expected from such as simple dependence on distance, thus underscoring the fact that the light pollution is still a significant factor at those relatively large distances. This fact is even more evident from the measured irradiance in the 443-nm wavelength channel, in which the values measured at stations i10 and i19 at a depth of 1 m under the water's surface (8 × 10 −5 μ W cm −2 nm −1 and 7.6 × 10 −5 μ W cm −2 nm −1 , respectively) are almost as high as the value measured at station i1 at the same depth (1.3 × 10 −4 μ W cm −2 nm −1 ). In addition to the superposition of multiple light sources, sky glow caused by scattering of the light in the atmosphere, as mentioned in the Introduction, can also distribute light from the coastal light sources out to the middle of the gulf.
In Fig. 5, a vertical profile of downwelling irradiance at station i4 in three wavelength channels [520 nm (green light), 589 nm (yellow light), and 443 nm (blue light)] is shown. From Fig. 5, one can see that the intensity of the lights of the city of Eilat and its industrial sections, especially in the yellow and blue parts of the emitted light spectrum, is high enough for the light to penetrate beyond the first few meters of the water column. A weak signal (close to the measurement threshold) was even detected at a depth of 30 m. However, at the more remote measuring locations, the signal to noise ratio dropped within a few centimeters below the water surface. As is evident in Fig. 5, at station i4, down to a depth of ~7 m, the irradiance of yellow light was higher than that of blue light, but below a depth of ~7 m, the irradiance of blue light became higher than that of yellow light. The higher rate of    . 2). Ed ranged from 4.6 × 10 −4 to 1 × 10 −6 μ W cm −2 nm −1 in the 589-nm (yellow) channel, from 7 × 10 −5 to 1 × 10 −6 μ W cm −2 nm −1 in the 520-nm (green) channel, and from 1.3 × 10 −4 to 5.8 × 10 −5 μ W cm −2 nm −1 in the 443-nm (blue) channel. The irradiance of green and yellow light reached the instrument limit at depths of 15 and 20-m, respectively. The irradiance of blue light did not reach noise level down to the maximum depth to which the instrument was deployed (30 m).
decrease of the irradiance of yellow light with depth is due to the correspondingly higher attenuation coefficient of yellow light in such waters 62 (case I waters; refer to the Introduction) (e.g., K d at 443 nm = 0.035 m −1 ; K d at 589 nm = 0.14 m −1 56 ), as well as due to the geometrical configuration of the measurements. (At the air-water interface, yellow light is refracted into a direction farther from vertical than blue light, and therefore the yellow light propagates along a longer path from measurement depth to measurement depth under the water. The fact that the sources of light are localized/pseudo-point sources, and that fact that at each station we measured irradiance through a localized area parallel to the water's surface, also adds a factor of a decrease with distance squared to the cosine of the local zenith angle, which only accentuates this same effect. For example, from a rough estimation of the attenuation coefficient based on a combination of our daytime and nighttime irradiance measurements (not shown here), the attenuation coefficient of yellow light is ~0.05 m −1 higher at depths of ~5-20 m under the water due to geometrical effects than it would be for a plane parallel light source, while the attenuation coefficient of green light is only ~0.03 m −1 higher than it would be for a plane parallel light source and only at depths of ~5-10 m under the water, and the attenuation coefficient of blue light is at most 0.01 m −1 higher than it would be for a plane parallel light source and only over a range of a few meters depth).
In Fig. 6, the spectrum of downwelling irradiance on the water surface on the night of August 12 th , 2015 at 22:00 local time is shown for two locations, a more illuminated station (station i1 in Fig. 2; black curve) and a less illuminated station (station i19 in Fig. 2; gray curve). From Fig. 6, one can see that not only does the absolute value of irradiance differ at the two stations but also the spectral dependence. The peak in the 589-nm wavelength channel due to sodium lighting is especially prominent at station i4, close to the city and the port. In contrast, at the less illuminated station, the blue (443 nm and 465 nm) parts of the spectrum exhibit higher peaks, due to the different types of artificial lighting used in proximity to station i19. Such spectral information at different sites in the gulf is important for assessing the extent and type of impact of the light pollution on the local environment.
Also shown in Fig. 6 are spectral measurements of moonlight on the night of a full moon (the night of September 9 th , 2014) at 23:00 local time. From Fig. 6, one can see that at station i4 (again near the city and port) on a moonless night, the measured irradiance in the 443-nm and 465-nm wavelength channels are of comparable values to the measured irradiance on the night of the full moon. Furthermore, the irradiance measured in the 589-nm wavelength channel at station i4 on the moonless night is higher than the irradiance measured in the 589-nm channel at station i19 on the night of a full moon. Therefore, artificial nighttime illumination in the Gulf of Aqaba is indeed comparable to and may even exceed the illumination of moonlight. Note that since the intrinsic attenuation coefficient of the water would be the same for both artificial nighttime light and moonlight, and since, as we see from Fig. 6, the irradiance of artificial nighttime light is similar to or larger than the irradiance of moonlight at the surface of the water, the same would be true deeper in the water column where the corals are located.
Our measurements show a clear gradient of unnatural illumination originating from the cities of Eilat and Aqaba and their surroundings into the waters of the Gulf of Aqaba. We have found that in certain wavelength channels, there is almost a two orders of magnitude difference in the irradiance of nighttime light pollution in the Gulf of Aqaba between reef areas next to Eilat and sections of the gulf that are more distant, both at the surface and under water. The fact that the irradiance of artificial light is not uniform over the water body indicates that the effect on water biota may not be uniform. We note that while in terms of vision, a gradient of two orders of magnitude is not extreme (marine animals that navigate vertically through the water column during the day experience similar or larger gradients of light), it is a gradient of light that exists when no light should exist. The light pollution we have measured in the current study is a potentially harmful factor due to the unique proximity of the coral reefs to the shore and to the urban light sources. The unique geographic setting has to be considered in relation to the manifold different aspects of the influence of light on the structure and functions of coral reefs 33,[63][64][65] . This underscores the importance of mapping the spatial and spectral distribution of artificial light in area as Gulf of Aqaba and determining the gradients of light pollution along the gulf. Such mapping provides an experimental framework for the rigorous study of the manifold effects of light pollution on marine life, enabling a comparison of exposed reef sectors to more distant reef sectors as controls. Furthermore, determining the spectral characteristics and sensitivity thresholds of light induced damages and perturbations will provide city planners and municipal lighting designers with critical information necessary to prevent further damage to the gulf 's most important natural treasures and the main socioeconomic resources of the region's residents.
With respect to the effect of artificial nighttime light on phototroph photosynthesis, Raven and Cockell 26 measured PAR close to a city and estimated that the combination of full moon light and sky glow (including the contributions of scattering by water vapor and reflection by clouds) could in some cases reach the lower limit required for photosynthesis (0.1 μ mol photons m −2 s −1 66 ). At the water surface in the most illuminated locations, we measured a value of total PAR that is slightly above 0.1 μ mol photons m −2 s −1 , namely ~0.5 μ mol photons m −2 s −1 . Nevertheless, according to our estimation, this total PAR is likely to have a negligible effect on net carbon fixation. [Compare these values to the typical PAR of ~100-2,000 μ mol photons m −2 s −1 that we measured under daylight conditions between the hours of 11:00 and 13:00 local time on a monthly basis from August 2014 to December 2015 using the same methods (results not shown here)]. Under the water surface and at less illuminated locations, where the total PAR measured is even lower, photosynthesis should be completely negligible.
Though the effect of the total PAR on net carbon fixation and photosynthesis is likely to be negligible, the light pollution at the levels reported in the current study may still affect more sensitive physiological and behavioral patterns and processes. These might include synchronization of biological clocks, navigation, migration, reproduction, phototaxis and bioluminescence, and may thereby lead to the impairment of regional marine organisms, as suggested elsewhere (e.g., refs 16 and 67). For example, as we mentioned in the Introduction, several species of corals have been found to be extremely sensitive to the blue region of the light spectrum 10,21,24 . Moreover, the fact that the irradiance of artificial light at certain stations and in certain wavelengths is similar to or even higher than the irradiance of moonlight on full moon nights (Fig. 6) reinforces the potential disruptive influence of light pollution on the marine ecosystems in the gulf. Therefore, we suggest that future studies of the effect of blue light and of light pollution in general on the specific biological functions of marine organisms and of their ecosystem-level consequences in the Gulf of Aqaba should be given high priority.

Conclusions
From our in situ measurements, we have found that nighttime illumination in the Gulf of Aqaba is dominated by the spectral characteristics of low pressure sodium vapor lights. However, we found that the 589-nm sodium vapor light signature at the surface becomes secondary to the irradiance of blue (443 nm) light from a depth of ~7 m; we found that this 443-nm wavelength light is the most penetrating wavelength in the waters of the gulf. Therefore, the ongoing change in the type of lighting used (the tendency to use more LED lighting, with a stronger blue component) will likely result in a considerable increase in the amount of light that will reach deeper into the water column and to the sea floor, thus exposing more areas of the reef to nighttime light.
We have found that in some locations, the irradiance of near shore artificial light is equal to or exceeds the irradiance of the light of the full moon. Hence, we expect that nighttime light pollution in those locations will interfere significantly with the activities of marine organisms that are synchronized with the phases of the moon. There is a particular danger of upsetting diurnal light/dark vision based behavioral feeding and feeding avoidance patterns of reef dwelling and pelagic organisms, which would destabilize the ecosystem structure and its functions.
Researchers currently face the challenge of disentangling the cumulative effects of all of the facets of human disturbance on coastal ecosystems with which artificial night lighting is often correlated, such as urban development, noise, exotic invading species, animal harvesting, and resource extraction. Even assessing the separate and combined impacts of direct artificial light and sky glow is not trivial.
By providing researchers and decision makers with hitherto unavailable information about the spatial dispersion of disturbed and relatively pristine reef sections of the Gulf of Aqaba, the data presented in the current study should serve as an important contribution towards further research on light pollution and coral reefs and towards future legislation. We expect that this study will prompt follow-up studies on the effects of light pollution on diverse coastal, shallow water ecosystems, helping to predict future shifts in their structure, assemblages, and function. We also expect that our results will promote future efforts to combine an assessment of the influences of light pollution on the specific physiology and behavior of coral reefs and their denizens with an assessment of the bathymetric and geographic boundaries of such effects. Locally, we expect that this study will serve scientists, governmental and local authorities, and interested NGOs in their efforts to ensure the health of the Gulf of Aqaba's unique coral reefs. The study will have immediate effects on the design of municipal lighting systems and related legislation. We suggest that as coastal cities such as Aqaba and Eilat continue to develop, measurements of light pollution should be included routinely as part of environmental monitoring protocols on both sides of the gulf, and we hope that this study will provide an impetus for collaboration between scientists in Jordan and Israel on this important topic. We also hope that the unique data presented here will raise general public awareness of the issue of coastal light pollution both locally and globally.