Wavelength-dependent effects of artificial light at night on phytoplankton growth and community structure

Abstract

Artificial light at night (ALAN) is a disruptive form of pollution, impacting physiological and behavioural processes that may scale up to population and community levels. Evidence from terrestrial habitats show that the severity and type of impact depend on the wavelength and intensity of ALAN; however, research on marine organisms is still limited. Here, we experimentally investigated the effect of different ALAN colours on marine primary producers. We tested the effect of green (525 nm), red (624 nm) and broad-spectrum white LED ALAN, compared to a dark control, on the green microalgae Tetraselmis suecica and a diatom assemblage. We show that green ALAN boosted chlorophyll production and abundance in T. suecica. All ALAN wavelengths affected assemblage biomass and diversity, with red and green ALAN having the strongest effects, leading to higher overall abundance and selective dominance of specific diatom species, some known to cause harmful algal blooms. Our findings show that green and red ALAN should be used with caution as alternative LED colours in coastal areas, where there might be a need to strike a balance between the effects of green and red light on marine primary producers with the benefit they appear to bring to other organisms.

1. Introduction

During the last century, the use of artificial light at night (ALAN) has increased considerably. Recent analyses have suggested that ALAN, which is strongly associated with the increasing worldwide urbanization [1], is still currently spreading spatially at a rate between 2% and 6% per year [2,3], with a parallel increase in irradiance at 2.2% per year [3]. The surge of ALAN has altered natural lightscapes, which in turn may have dramatic effects on wild species and ecosystems [4]. Indeed, the impact of ALAN on wildlife and ecosystems has received a lot of attention in the last two decades [5–8]. In vertebrates, ALAN has been linked to several behavioural and physiological effects, such as disruption of circadian rhythms [9], altered reproductive timing [7,10], poor sleep [11,12], reduced immune function [13] and altered metabolism [14]. Insects are also heavily affected [15–17], particularly because of the strong phototaxis found in many species [18,19].

Despite the surge of interest in the ecological effects of ALAN, most of the evidence collected so far comes from terrestrial habitats, while studies on marine populations and communities are currently limited [20–25]. Although a small number of studies have investigated the effects of monochromatic ALAN on cyanobacteria and microalgae, these focused on benthos and periphyton [2,26–30], while no study so far has investigated the effects of polychromatic LEDs on marine phytoplankton. Due to the continuous expansion of coastal urbanization [31,32], ALAN is a source of pollution that is increasingly relevant for coastal ecosystems [4,21]. Coastal ecosystems globally are also increasingly affected by eutrophication and harmful algal blooms (HABs) due to nutrient-rich inflows from either agricultural or urban sources [33,34]. Given the ecological importance of light for photoautotrophic phytoplankton species, the potential severity of ripple effects from phytoplankton to higher trophic levels [29] and the existing vulnerability of coastal systems to eutrophication, it is imperative to determine the type and magnitude of the response of marine primary producers to ALAN.

In primary producers, light is a strong modulator of photosynthesis and associated processes driving growth and cell fitness [35]. Light absorption by chlorophyll peaks at approximately 430 nm, although a second, lower peak is also present at longer wavelengths of 670 nm [27,36,37]. Moreover, light serves as an informational cue, regulating the synchronization of diverse intracellular processes ranging from phototactic, photoprotective and physiological responses essential for growth and development [38–40]. For example, green algae have an ‘eyespot’ with which, through a rhodopsin-mediated signalling pathway that is sensitive to green light, are able to direct their movement [41,42]. Therefore, the disruption of light cycles by ALAN has the potential to impact organism physiology and consequently assemblage structure via multiple pathways that are responsive to different light wavelengths. This is a topical question because the spectral composition of ALAN is also changing along with its surge in intensity, since many countries are replacing traditional lighting sources with the cost-efficient, energy-saving light-emitting diode (LED) technology [8]. LEDs are very flexible light sources whose colour can be easily modified. Indeed, new light installations use LEDs of different colours. While cool white LEDs (richer in blue/green wavelengths) are the most widespread, warm white (rich in yellow/orange/red wavelengths), green and red LEDs are also in use [8,21].

Recent findings have demonstrated that ALAN from warm white high-pressure sodium lamps can affect multiple signalling events and metabolic pathways essential for photosynthesis in freshwater cyanobacteria [43]. ALAN from cool white LEDs can increase the photosynthetic biomass of microphytobenthos [29] and its temporal variability [30], alter periphyton composition [44] and modify the community structure of freshwater benthic microorganisms [26]. However, the aforementioned studies experimented with a single ALAN wavelength, while it is increasingly recognized that different wavelengths can cause profoundly different responses in wild organisms [45–51]. This can potentially lead to competing conservation goals [51]. Compared to the open ocean where short (blue) wavelengths propagate best, in coastal waters, green-yellow wavelengths, including those produced by ALAN sources [52] are more dominant [53,54]. In coastal areas, red LED light has been recommended as a source of illumination because it doesn't interfere with sea turtle nesting and hatching [55], as well as with coral biology [56], whereas green light was suggested to minimize the impact of ALAN on seabird navigation [57]. To understand and therefore inform the ecological management of coastal areas, it is also essential to establish the effects of different wavelengths of ALAN on important aspects of microalgae assemblages, ranging from single-species growth to community-level properties such as diversity and species composition.

In the present study, we investigate experimentally the response of marine phytoplankton to three ALAN wavelengths (white 4500 K, green 525 nm, red 624 nm) compared to dark nights. Our first objective is to determine whether different wavelengths of ALAN can have a different impact on the growth of a single phytoplankton species. Furthermore, we aim to assess whether different wavelengths can have different impacts on the diversity and species composition of a phytoplankton assemblage. We hypothesize that white ALAN might stimulate growth compared to the dark, as it partly overlaps in wavelength with the first absorption peak of chlorophyll-A, an abundant pigment in all microalgae, at approximately 465 nm [36]. Conversely, we predict that the green and red ALAN should have a weaker, if any, effect as its spectral properties have minimal overlap with the light absorption range of chlorophyll-A (electronic supplementary material, figure S1). Finally, we predict that the effect on single-species growth could cascade to the community level, as phytoplankton species' competitive ability has been shown to shift with water colour [36].

2. Material and methods

(a) . Experimental set-up and light sources

To test the effect of ALAN wavelengths on single-species growth and assemblage biomass and diversity, we run two concurrent experiments from 23 November 2020 and for a period of 18 days: experiment 1 tested the effects of ALAN on the green microalgae Tetraselmis suecica and experiment 2 on a natural coastal assemblage dominated by diatom species. The experimental design comprised four treatments: control (12 L : 12 D), and three different ALAN wavelengths, green, red and white (12 : 12 light–ALAN). Each of the two experiments comprised five replicated cultures within each of the four treatments for a total of 40 experimental units (Erlenmeyer flasks of 200 ml each).

Since algae use light as a source of both information and energy [58,59], light treatments were standardized to levels of irradiance (i.e. energy content) rather than illuminance (i.e. luminous flux incident on a surface). Daytime light irradiance was 6.5 W m⁻² and was provided by a 10 W flood light (Prolite, Ritelite Systems Ltd, UK) equipped with two arrays of high-power LEDs (6000 K). Each ALAN source consisted of a strip of three LED diodes. The green ALAN wavelength was 525 nm (MULTICOMP), the red was 624 nm (MULTICOMP) and the broad-spectrum white LED light contained a higher peak at 470 nm and lower peaks between 550 and 600 nm (BROADCOM) (for full spectral characteristics of LED lights see electronic supplementary material, figure S1). The emission spectra were measured by a spectrometer (AvaSpec-2048 L, Avantes, Apeldoorn, The Netherlands). Night-time light irradiance was measured with a LI-200R pyranometer (LI-COR, USA) and was standardized at 0.023 W m⁻² for all three ALAN treatments. This irradiance level is within the range of values reported in previous studies on ALAN [26,28,60,61]. With respect to illuminance, our standardized irradiance level corresponds to 8.51 lux for the white light, to 3.4 lux for the red light and 12.81 lux for the green light. These values are ecologically relevant and within plausible ranges of ALAN observed in near shore epifaunal invertebrate assemblages (0.005–21.6 lux) [23,62,63].

The distance between the surface of the water in the flasks and the LED lights was approximately 40 cm. Each light treatment was applied inside a light-proof box (55 × 62 × 62 cm), where the experimental replicates were introduced (see below for details on how these were produced in each experiment). The replicates were partially submerged (by one-third of the flask height) into water baths (44 × 41 × 22 cm). Light irradiance was measured at mid flask height at six different locations within each water bath and was not statistically different between the ALAN treatments both during the day (linear mixed model, χ = 3.9, p = 0.41) and during the night (linear mixed model, χ = 4.1, p = 0.13). The water baths were maintained at a temperature of 14–15°C chosen to reflect the mean annual sea surface temperature of mid-latitude seas. To minimize box effects, water in the baths had an identical temperature as it was fed from a central tank where the temperature was regulated. Cultures were mixed once a day when their position inside each treatment box was also randomized.

(b) . Experimental procedure for experiment 1: single-species response

The green microalgae T. suecica was selected for the single-species response experiment because of its use as a model species in studies using continuous illumination with monochromatic LEDs [64–66], because of the industrial potential of the species as a high-lipid content strain [67], and its importance as fish and shellfish aquaculture feed [68].

Inoculum from our T. suecica culture (sourced by CCAP 66/4) was grown in F/2 medium (Guillard 1975) made by ultrapure artificial seawater at 35 ppm salinity (V = 200 ml). All cultures were initiated at a concentration of 5000 cells ml⁻¹. Every second day, 5 ml samples were taken from each replicated culture, 2 h after the onset of day light in the morning, to calculate cell numbers and growth rate. Cells were counted using Fast-Read 102 counting chambers under a light microscope. T. suecica growth showed a lag phase of 8 days due to the acclimation of cells from 20°C to 15°C and thereafter growth entered the exponential phase. Maximum growth rate for each replicate culture was determined based on the formula μ = ln(N₂/N₁)/(t₂ – t₁), where μ is the specific growth rate, and N1 and N2 are the cell number at time 1 (day 8) and time 2 (day 18), respectively. On day 18 of the experiment, 50 ml samples were also taken to determine chlorophyll-A concentration according to Parsons et al. [69].

(c) . Experimental procedure for experiment 2: diatom assemblage response

We used a natural diatom-dominated marine sample in our assemblage response experiment. Specifically, we assorted at equal volume (200 ml) across the 20 experimental replicates, an inoculum of unfiltered marine surface water collected at 50 cm depth from the shore of Largs, Scotland (55.794659, −4867615) on 22 November 2020, 11:00. The initial inoculum had a chlorophyll-A concentration of 0.7 µg l⁻¹ and salinity 30 psu which is lower than the salinity of the open sea as the area receives freshwater inflows.

The culture medium consisted of the collected marine sample and added nutrients commensurate with F/2 medium concentration [70]. Species identities, cell counts and chlorophyll concentration were determined on day 12 of the experiment when cultures just entered the stationary phase as determined by the cell counts of selected replicates. Specifically, a 5 ml sample was collected for species identification and was preserved with Lugol's iodine solution. Samples were subsequently filtered through a Sartorius cellulose nitrate membrane filters (0.45 µm pore size, 25 mm diameter) and dried in an incubator at 40°C for an hour. The filter was made transparent by the addition of a drop of immersion oil and was observed under a light microscope (40×/0.65) where 15 randomly selected fields of view were used to identify and enumerate the different species. The volume of sample examined was equal across all samples thus species' cell counts as well as total assemblage cell counts are directly comparable across replicates and reported as counts. Chlorophyll concentration was determined from 50 ml samples as in the case of the T. suecica experiment.

(d) . Data analysis

For the T. suecica single-species response, we used three Gaussian linear models to determine the effect of ALAN treatment (four levels: green, red and white ALAN and the dark control) on each of three response variables: the growth rate, and the cell number and chlorophyll-A measured on the final day of the experiment (day 18).

For the diatom assemblage response, we used four linear models to test the effect of ALAN treatment on each of four response variables: assemblage total cell count, chlorophyll-A, Menhinick richness [71] and Pielou's evenness [72]. The Menhinick species richness index is defined as the number of species in the sample divided by the square root of the total abundance of individuals in the sample and was used to enable standardization of species richness across samples based on the total cell abundance. Pielou's evenness index is defined as the Shannon diversity divided by the maximum possible value of Shannon (if all species had equal abundance in the sample) and was used to provide a measure of dominance in cell counts by specific species in a sample. The Menhinick and Evenness indices were sensitive in expressing changes in phytoplankton diversity in previous studies comparing multiple diversity indices using phytoplankton species-abundance data [73]. Assemblage total cell count, chlorophyll-A and Menhinick richness were modelled with Gaussian models. Evenness was modelled with a beta distribution model as its values were confined between 0 and 1 and the effect of treatment was tested using likelihood ratio test (LRT) between the null model (not containing treatment) and the full model (which contained the factor treatment).

To test the effect of ALAN treatment on assemblage composition, we performed the analysis of the similarity between all pairwise combinations of the 20 replicates using the Bray–Curtis similarity index [74] on non-transformed species-abundance data. We visualized these similarities using cluster analysis to check the grouping of samples based on the different treatments. We fitted additional linear models to test for the effect of treatment on the abundance of specific diatom species. Finally, the percentage changes we report in the first paragraph of the discussion (e.g. for the cell number) were calculated according to the formula: [(cell number of ALAN treatment - cell number of dark control)/cell number of dark control] × 100.

Model selection was carried out based on the least-squares approach apart from the evenness model where we used the LRT test. We also conducted post hoc pairwise t-tests to assess differences between the four treatment levels. All statistical analyses were carried out in R v. 3.5.0 (RStudio Team, 2016). The packages ggplot2 v. 3.3.0 [75], ggpubr v. 0.2.5 [76], ggdendro v. 0.1-20 [77] and dendextend v. 1.13.4 [78] were employed for plot generation and data visualization. The package emmeans was used for pairwise comparisons between treatment levels [79]. For data manipulation, reshape2 v. 1.4.3 [80], plyr v. 1.8.6 [81]. For modelling the beta distribution, we used the function glmmTMB and family function beta_family(link = ’logit’) in the package glmmTMB [82]. We used the vegan v. 2.5-6 R [83] and cluster v. 2.1.0 packages [84] to perform the pairwise similarity of species-abundance data and related cluster analysis.

3. Results

(a) . Green and red artificial light at night promote growth of the green microalgae Tetraselmis suecica

The ALAN treatments had a statistically significant effect on the growth rate of T. suecica cells (F_3,16 = 6.64, p = 0.004), and this was shown to be wavelength specific. Specifically, a significantly higher growth rate was observed under the green ALAN treatment compared to white ALAN and the dark treatment. Furthermore, the growth rate response to red ALAN was also higher than the white ALAN treatment (figure 1a; electronic supplementary material, table S1).

Figure 1.

ALAN affects growth rate, cell and chlorophyll-A concentration of Tetraselmis suecica in a colour-dependent manner. Effect of ALAN treatments (dark, green, red and white) on the growth rate calculated during the exponential growth phase (days 8–18) (a), on the cell concentration at day 18 (b) and chlorophyll-a concentration at day 18 (c). Pairwise comparisons show differences between treatments (not shown: p > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001). (Online version in colour.)

The T. suecica cell concentration was significantly affected by the ALAN treatment (F_3,16 = 7.691, p > 0.002). Specifically, on day 18 of the experiment, the cell number was significantly higher in response to the green ALAN treatment compared to the white ALAN and dark treatments and was also higher in the red ALAN compared to white. No difference was observed between the white and dark treatments (figure 1b; electronic supplementary material, table S1). These results are comparable to those obtained in a pilot experiment where LED colours were allocated to different experimental boxes and light intensity was standardized at 20 lux (for details of this pilot experiment and relative results, see electronic supplementary material, figure S2).

The chlorophyll-A concentration of Tetraselmis cultures on day 18 was also significantly affected by the ALAN wavelength (F_3,16 = 20.584, p > 0.001). Specifically, chlorophyll-A content was significantly higher in the red and green ALAN treatments compared to the dark and white treatments, whereas no difference was observed between the white and dark treatments (figure 1c; electronic supplementary material, table S1).

(b) . Artificial light at night affects assemblage biomass and diversity

Analysis of the initial inoculum upon collection from the sea showed that the assemblage comprised 14 species of which 11 species were Diatomophyceae, two were Dinophyceae and one was Dictyochophyceae. The most dominant species were Skeletonema sp. (28% dominance), Cyclotella sp.1 (19% dominance), Cyclotella sp.2 (17% dominance), Ceratium lineatum (15% dominance) and Navicula sp.1 (6% dominance) whereas all other species were subdominant with relative abundance less than 2%. On day 12 of the experiment, the overall biomass had considerably increased and stabilized across treatments and assemblages. Experimental units on day 12 comprised 4–7 species of diatoms (13 species overall across all treatments). The planktic colonial diatoms Skeletonema sp., Thalassiosira nordenskioeldii and to a lesser extend T. eccentrica were more dominant across treatments. However, the absolute and relative abundance of Skeletonema sp., T. nordenskioeldii presented differences between treatments as discussed below.

ALAN affected assemblage cell counts and diversity independent of colour whereas chlorophyll-A concentration was affected in a wavelength-specific manner. Specifically, the total diatom assemblage cell count (i.e. cells summed across all species in the assemblage) was significantly affected by every ALAN treatment tested (F_3,16 = 9.589, p > 0.001). Cell count was statistically higher under all ALAN wavelength conditions compared to the dark but no differences were observed between the ALAN wavelengths (figure 2a; electronic supplementary material, table S2). The chlorophyll-A concentration of the diatom assemblage was also significantly affected by the variable treatment (F_3,16 = 12.393, p > 0.001), with all ALAN wavelengths having a higher concentration compared to the dark control whereas the red wavelength was also higher from the green and white (figure 2b; electronic supplementary material, table S2). A significant effect was also observed on the assemblage evenness (LRT, d.f. = 3, p < 0.001), whereby the assemblages under all ALAN wavelengths had significantly higher evenness (more evenly distributed species' populations) compared to the dark control (figure 2c; electronic supplementary material, table S2 for results of post hoc tests). A significant effect of ALAN was observed on the Menhinick richness (F_3,16 = 3.260, p = 0.049), with the dark treatment showing significantly higher richness compared to all ALAN wavelengths tested (figure 2d; electronic supplementary material, table S2).

Figure 2.

ALAN affects assemblage biomass and diversity. Effect of treatment (dark, green, red and white) on the total cell counts (a), chlorophyll-A (b), evenness (c) and Menhinick richness (d) measured on day 12 of the diatom assemblage experiment. Pairwise comparisons show differences between treatments (not shown: p > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001). (Online version in colour.)

(c) . Artificial light at night affects species’ relative abundance

ALAN did not lead to a shift in the species' identities comprising the assemblages. However, ALAN affected the absolute and relative abundance of species (measured as standardized cell counts) within treatments in a wavelength-specific manner. In particular, assemblages growing under the green and red ALAN were 80% similar and were 44% dissimilar from the assemblages under the dark control and white ALAN conditions, respectively, with the exception of two replicates which grouped with the dark/white cluster mainly due to lower numbers in Skeletonema sp. (figure 3a). This was due to a significant increase in the species Skeletonema sp. and T. nordenskioeldii relative to the subdominant species in the assemblage (i.e. all species excluding Skeletonema sp., T. nordenskioeldii and T. eccentrica). Specifically, Skeletonema sp. had a significantly higher abundance in all ALAN colours compared to the control (F_3,16 = 7.708, p = 0.002) (figure 3b). T. nordenskioeldii was significantly higher in response to the red ALAN treatment compared to the white and the dark control (F_3,16 = 8.8574, p = 0.001) (figure 3c). No differences between the treatments were observed in the abundance of the subdominant species (F_3,16 = 0.833, p = 0.495) (figure 3d). These differences in relative abundances between dominant and subdominant species (figure 3b–d) suggest that the increased evenness in ALAN treatments compared to the dark control (figure 2c) was likely due to the increased evenness of the dominant species in the assemblage (figure 3b,c) rather than an increase in evenness across dominant and subdominant species.

Figure 3.

Red and green ALAN lead to similar responses in assemblage composition. Cluster showing the pairwise similarities between the replicate assemblages based on the Bray–Curtis similarity index calculated on non-transformed species-abundance data (a). Pairwise comparisons of abundances of Skeletonema sp. (b), Thalassiosira nordenskioeldii (c), the sum of all subdominant species in the assemblage (d) (not shown: p > 0.05, *: p ≤ 0.05, **: p ≤ 0.01, ***: p ≤ 0.001). (Online version in colour.)

4. Discussion

In this study, we tested the effect of different ALAN wavelengths on phytoplankton growth, assemblage diversity and species composition. We predicted that the effect on single species and assemblage level would be more pronounced under white ALAN, compared to dark and red and green ALAN, as it partly overlaps in wavelength with the first absorption peak of chlorophyll-A, an abundant pigment in all microalgae [36]. Contrary to our expectations, our findings suggest that red and green ALAN have a more pervasive impact on phytoplankton growth and assemblage structure compared to the white ALAN. More specifically, our experiments showed that exposure of the green microalgae Tetraselmis suecica to green ALAN led to a 5% increase in growth rate, 67% in cell number and 49% in chlorophyll-A concentration, compared to the dark night condition. Exposure to red ALAN led to a similar response to the green ALAN, as it resulted in higher chlorophyll-A, but it did not affect growth rate and total cell number compared to the dark treatment. Red and green ALAN treatments also affected the diatom-dominated phytoplankton assemblage. For example, red ALAN led to higher total cell count and chlorophyll-A concentration by 118% and 80%, respectively, compared to the dark control. More interestingly, red and green ALAN led to a similar assemblage response by balancing the biomass of the most abundant species (thus leading to higher evenness), but also by enhancing the biomass of the most abundant species relative to the subdominant species. These effects were less pronounced in response to the white ALAN treatment although it had a significant impact on assemblage richness.

Previous studies on the effect of white ALAN on freshwater primary producers have reported longer term (six-week experiments) increases in the abundance of benthic microalgae [26] but also shorter term (three weeks) decreases in periphyton abundance, as well as community composition shifts [27]. Our experimental findings provide additional insights into ALAN effects by offering comparative information on different LED colours. Our findings show that although exposure to white ALAN can lead to changes in diatom assemblage diversity, species’ relative abundances and biomass increase within 12 days of exposure, this effect was less pronounced compared to the red and green LED, and our white ALAN treatment had no effect on the growth rate of the green microalgae Tetraselmis.

Our findings also suggest that the red and green ALAN colours have the potential to enhance the growth of HAB species such as the diatom Skeletonema sp. This species is commonly known for forming dense blooms causing mortality to other organisms through physical damage (e.g. fish gill lesions) or anoxia, consequently impacting trophic interactions, biodiversity and overall ecosystem health [85]. This finding is supported by Oh et al. [86] who showed that green LEDs can selectively stimulate the growth of the diatom Skeletonema costatum compared to other species in the assemblage. In addition, the biomass of planktic colonial diatoms such as Skeletonema and Thalassiosira was enhanced under red and green ALAN compared to epiphytic (i.e. growing on macroalgae and rocks) and epipsamic (i.e. growing on sand) diatom genera. Given that maximal transmission of light in coastal systems is around 550 nm (green/yellow) [53] and that coastal seafloors are susceptible to ALAN, particularly within the green range (495–560) [52], we could anticipate impacts of ALAN on phytoplankton biomass and assemblage structure in coastal ecosystems.

A key question is why green and to some extent also red ALAN had a stronger effect on the growth and photosynthetic biomass of Tetraselmis compared to white light. A first insight stems from comparisons with previous studies that focused on maximizing the growth and biochemical composition of Tetraselmis to fully exploit the industrial potential of this algae. Unlike our 12 L : 12 D period, these studies used continuous (24 h) high intensity LED illumination of different monochromatic LEDs [64–66]. Abiusi et al. [64] reported maximum growth and biomass concentration under red and white continuous light, whereas these traits were less pronounced under green light (all light conditions were standardized at 160 µmol m⁻² s⁻¹). Aidar et al. [65] also reported increased growth under continuous red and white light compared to the blue-green light (all light conditions were standardized at 25 μmol m⁻² s⁻¹). These results contrast with the higher growth rate under green, dim night-time illumination found in our experiment. This discrepancy raises the question of whether dim ALAN has the potential to induce different responses to LED wavelengths compared to higher intensity light at night.

Stimulated growth and photosynthetic activity in our green algae monoculture and diatom assemblage under the red light could be explained by the partial overlap of chlorophyll-A absorption spectra with the red emission range [87] (see also electronic supplementary material figure S1). The absorption of other photosynthetic pigments present in diatoms (chlorophyll-C, β-carotene, zeaxanthin, diatoxanthin, diadinoxanthin and fucoxanthin) and green algae (chlorophyll-B, β-carotene, zeaxanthin, violaxanthin, neoxanthin and loroxanthin) [66] typically peak in the 420–480 nm range (see datasets from [78]) and although they partially overlap with our green ALAN spectrum this does not justify why higher growth and biomass was not also observed also in the white ALAN treatment with which they overlap significantly more (see electronic supplementary material figure S1 for our ALAN light spectra).

Although pigments such as chlorophylls within chloroplasts absorb light energy to fix inorganic CO₂ towards biomass production and growth, other photopigments act as photoreceptors and are involved in functions that regulate circadian rhythms and phototaxis [54,59]. Some of these, such as phytochromes, are sensitive to red/far red wavelengths of light. One conceivable hypothesis to explain increased growth under red and green, but not white ALAN, is that the white ALAN has disrupted the natural photocycle of Tetraselmis sp., with downstream consequences on photosynthesis, cell division and growth [59,88,89]. This hypothesis could be tested by simultaneously monitoring chlorophyll, growth and clock gene expression under different ALAN colours.

Nevertheless, the strong effects of green ALAN on the abundance and photosynthetic biomass seen in our study are still puzzling considering that a green light receptor has never been found and that cell division is typically stimulated by blue light. A photoreceptor that may have played a role in our study could be rhodopsin, which is sensitive to light in the mid-range of the visible spectrum, peaking at approximately 500 nm. Rhodopsins are known from all algae groups and are associated with phototactic responses [54,90]. Although the activity of such photoreceptors can benefit microalgae growth in the marine environment where light is variable and often limiting, it is unclear how this mechanism led to growth stimulation in our controlled experiment where light conditions were more homogeneous. Nevertheless, this rhodopsin-mediated effect cannot be precluded since Tetraselmis is a flagellated microalgae and capable of movement to more optimal positions for capturing light (e.g. water surface in flasks). This would merit further testing with appropriate experiments that would track phototaxis in flagellated algae. Finally, although cryptochromes are the primary receptors of UV-A and blue light, it has been reported that green light affects cryptochrome photochemistry and activity as green light reverts cryptochromes to their inactive state [54,91]. In particular, cryptochromes integrate green light signals into the circadian system as well as modulating plant growth and architecture in response to an increase in green/blue light ratio under a canopy [92–94].

Our data show that there is a significant impact of green and red ALAN on phytoplankton that should be taken into account when planning nocturnal illumination in marine environments. In fact, these results may lead to conservation dilemmas, as both red and green LED lights have been suggested as alternative ALAN sources for public illumination. Specifically, red light illumination has been recommended in coastal areas because it interferes less with sea turtle nesting and hatching compared to broad-spectrum white light [55]. Similarly, the use of green light has been recommended to minimize the impact of light pollution on migratory birds [57]. In general, shifting spectral signatures toward longer wavelengths than blue light seems to be less harmful to many organisms, including insects [15], bats [95] and songbirds [96]. However, our study shows that the use of green and red ALAN LEDs can impact aquatic primary producers by enhancing the growth of different taxonomic groups (green algae and diatoms) indicating a potential to encourage eutrophication phenomena in marine coastal (but potentially also freshwater) systems where these taxonomic groups are also present. Although the batch culture set-up used in our study is more representative of coastal systems affected by pulsed nutrient inputs [97], it would be interesting to also simulate systems that show less pronounced fluctuations using continuous or semi-continuous nutrient supply set-ups.

The equivalent illuminance to the standardized irradiance used in our experiment ranged between 3.4 and 12.8 lux (depending on the colour). This is within the range of illuminance measured in coastal systems near ALAN affected areas by previous studies (0.005–21.6 lux) [23,62,63]. We thus conclude that effects on marine microalgae can be expected in coastal ecosystems and particularly in the proximity to shoreline illuminations, heavily urbanized environments and ports. In addition, given that both green algae and diatoms are also found in freshwater systems, we anticipate that our results may be relevant also for slowly moving riverine systems or lake systems also affected by ALAN.

Data accessibility

Our data are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.vdncjsxtn [98].

Authors' contributions

C.D.: conceptualization, investigation, methodology, visualization and writing-original draft; E.C.: conceptualization, funding acquisition, investigation, methodology, visualization, writing-review and editing; D.M.D.: conceptualization, investigation, methodology, resources, writing-review and editing; E.K.: conceptualization, resources, writing-review and editing; J.C.: methodology, writing-review and editing; N.M.: methodology, writing-review and editing; S.S.: conceptualization, formal analysis, investigation, methodology, resources, supervision, visualization, writing-review and editing

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper.

Funding

This work was funded by the A.G. Leventis Foundation and Nissad Development Company (sponsors had no further involvement in the research).