Spectral light quality on the seabed matters for macroalgal community composition at the extremities of light limitation

Compromised light quantity on the seabed affects the composition and resilience of marine forests and the maintenance of ecosystem services. However, the response of macroalgal assemblages to changes in the quality (spectral composition) of light as compared to the quantity alone is poorly understood. Here, we bridge radiative transfer modeling and algal physiology to simulate the underwater light environment and explore macroalgal responses to spectral attenuation by various coastal water constituents. We considered three groups of macroalgal taxa (class Phaeophyceae and, phyla Chlorophyta and Rhodophyta) for which we developed a “synthetic envelope” of spectrally resolved action spectra using 1 million radiative simulations. We found that brown algae were more efficient at harvesting light across the whole spectrum at low attenuation (< 0.2 m−1) when depths were greater than 10 m, whereas green algae dominated shallower depths than 10 m for low to high attenuation values. Finally, red algae made best use of the available light at relatively higher attenuation values (> 0.2 m−1) and deeper depths. We also demonstrated that the brown and green algal taxa showed higher photosynthetic efficiencies when attenuation was dominated by particulate matter rather than by phytoplankton or detrital matter. Our results shed new light on the “Complementary Chromatic Adaptation” hypothesis by confirming that algal pigmentation matters under a subset of scenarios, particularly at the limits of light availability in coastal waters. This study paves the way for identifying competitive thresholds in macroalgal communities by using satellite‐derived ocean color products to identify regions and depths of potential optical transitions.

Global biogeochemical cycling is driven by the conversion of electromagnetic radiation to organic matter via photosynthesis Field et al. 1998). Macroalgae are common in the world's coastal zone and their primary production is fundamental to the health of coastal ecosystems and to biogeochemical cycles (Gattuso et al. 2020;Duarte et al. 2022;Pessarrodona et al. 2022). Although photoautotrophs like macroalgae are frequently limited by other essential resources for photosynthesis and growth (Schimel et al. 1997), they are primarily limited in their depth and/or latitudinal distribution by light availability (Field et al. 1998;Gattuso et al. 2020;Duarte et al. 2022). In the coastal zone, benthic light availability (irradiance) for macroalgae is affected by the rate at which water and its dissolved and particulate constituents attenuate light through the water column, which simultaneously shapes the spectral composition of light reaching the benthos (Kirk 1994).
The "Complementary Chromatic Adaptation" (CCA) hypothesis postulated by the seminal work of Engelmann (1883) speculated that observed patterns of macroalgal zonation could be explained by the presence of several photosynthetic accessory pigments (e.g., phycocyanin and phycoerythrin) that favor the deepest penetrating wavelengths of light. Although the CCA hypothesis has been a focus of marine photosynthetic research for more than a century (Wurmser 1921;Klugh 1931;Haxo and Blinks 1950;Crossett et al. 1965;Larkum et al. 1967;Dring 1981;Ramus 1983;Lüning and Dring 1985;Morel 1991;Enríquez et al. 1994;Stomp et al. 2004;Stomp et al. 2007;Hintz 2021), it has been partially discredited and is not universally applicable (Crossett et al. 1965;Larkum et al. 1967), especially where light is not limiting (Dring 1981). Moreover, realized depth distributions are inevitably the result of variations in algal life history characteristics, nutrient availability, thermal performance, and phenotypic plasticity (Raven and Hurd 2012). Yet, when these complexities are accounted for using modeling approaches, accessory pigments theoretically provide a distinct advantage in light-limited scenarios in deeper water, particularly in more turbid conditions (Dring 1981). Although Dring (1981) rightly expresses the importance of light quantity in driving macroalgal distribution, it is important to recognize the myriad scenarios of spectral light distribution not covered by broad "water type" categories (Jerlov 1976) and the broad depth ranges we know that can be inhabited by macroalgae (Raven and Hurd 2012) that were not covered in the study of Dring (1981).
Here, we bridge optical modeling with algal physiology and ecology to model the competitive advantages inferred on red (phyla Rhodophyta), brown (class Phaeophyceae), and green (phyla Chlorophyta) macroalgae across broad depth and attenuation gradients. We expand on the optical modeling work by Dring (1981) by increasing the resolution of the optical conditions tested (nine water types and up to 10 m deep considered in the study of Dring 1981) as well as including more comprehensive action and absorption spectra collections from Enríquez et al. (1994) and Kirk (1994). Moreover, we use optical modeling to separate the relative consequences of two major coastal stressors, eutrophication, and sedimentation.
Most macroalgal assemblages exist in optically complex coastal waters (Gattuso et al. 2020;Duarte et al. 2022), where a range of land and ocean-derived organic and inorganic matter alter the absorption and backscatter of light and hence the spectral attenuation of light with depth. It is well known that a shifting light climate has the potential to greatly influence competitive dynamics of macroalgal assemblages with significant implications for net primary productivity, standing biomass, and rates of carbon sequestration (Clark et al. 2013;Desmond et al. 2015;Tait and Schiel 2018;Tait 2019;Blain et al. 2021;Smith et al. 2021;Weigel and Pfister 2021). For example, some coastal zones of New Zealand (NZ) experience high levels of sediment runoff (Schiel and Howard-Williams 2016), and these are associated with detrimental impacts to benthic macroalgal communities (Alestra et al. 2014;Tait 2019;Blain et al. 2020;Tait et al. 2021). Other NZ coastal zones have experienced regional shifts in phytoplankton communities driven by nitrogen loading (Safi et al. 2022). The relative consequences of these processes on the light available to broad macroalgal groups is unknown but is fundamental to light harvesting and maximal realized depth thresholds. Eutrophication favors the proliferation of phytoplankton, while "sediment" in the water column such as non-algal particulate (NAP) and color dissolved organic matter (CDOM) depress the total quanta of light as well as compress the available spectra, attenuating blue colors more strongly (Kirk 1994).
Physiologically, variation in the relative magnitude of these stressors has the potential to influence the success of macroalgal groups based on their broad range of photosynthetic pigments (e.g., carotenoids, chlorophylls, and phycobiliproteins) and their ability to use light of different wavelengths (Raven and Hurd 2012). The chlorophyll pigments absorb light strongest at the blue and red wavelengths and are present in all groups of macroalgae, while carotenoids (also present across macroalgal groups) often perform an accessory role in photosynthesis and can enhance blue wavelength absorption by green macroalgae (Ramus 1981;Hanelt et al. 2003). Red algae (Rhodophyta) are equipped with several accessory pigments, collectively called phycobiliproteins (e.g., phycocyanin and phycoerythrin), which promote absorption of green and yellow wavelengths (Dring 1981). It is these pigments that provide a marked difference in the spectral harvesting of light by red algae and promote effective use of green and yellow wavelengths (Haxo and Blinks 1950). Although there are high costs of synthesizing these protein complexes relative to the chlorophyll family (Raven 1984), the deepest growing macroalgae observed are generally rhodophytes (Littler et al. 1985;Runcie et al. 2008). Despite this observation and the coincidental drop in blue and red wavelengths in most coastal conditions, the relevance of the CCA hypothesis has been largely discarded since Dring (1981) study.
Because there is greater attenuation from the optically active constituents (OACs, that is NAP, CDOM and chlorophyll a [Chl a]) in the same part of the light spectrum of green and brown macroalgae action spectra (blue, green, and red wavelengths), we hypothesize that an increase in light attenuation (i.e., decreased water clarity) from coastal sediment and phytoplankton will affect the photosynthetic performance and efficiency of brown and green more than red macroalgae (Hypothesis 1). In addition, we suggest that for a given attenuation level, brown and green algae are favored, compared to red algae when relatively spectrally neutral particulate matter dominates attenuation rather than spectrally selective phytoplankton (Hypothesis 2). Finally, we postulate that coastal sediments and phytoplankton not only reduce the amount of light on the seabed but also remove crucial colors in the light spectrum and favor red-dominated macroalgal assemblages at the extremities of light limitation (Hypothesis 3). Understanding how the major water column constituents influence the benthic light spectrum in relation to the action spectra of key macroalgal groups will provide valuable insights into the possible implications of both sedimentation and eutrophication in coastal environments.

Approach
In this study, we generated synthetic benthic light spectra for various sets of optically active constituents and different depths (0-50 m) that we combined with typical action spectra of benthic macroalgal taxa from the literature (Enríquez et al. 1994;Kirk 1994). The synthetic data were used to investigate the interactive effects of the diffuse attenuation coefficient of downward irradiance in the photosynthetically active radiation (PAR) range (K d [PAR], m À1 ), depth (m), as well as the absorption by phytoplankton (a ph , m À1 ), absorption by detrital matter (a det , m À1 ), and particulate backscatter (b bp , m À1 ) on the quality of light reaching the seabed and usable in photosynthesis for the three main macroalgal taxa.

Sea surface light and atmospheric conditions
The WMO/WRDC Wehrli Air Mass Zero (AM 0) solar spectral irradiance curve was used as the extra-terrestrial solar spectral irradiance distribution (Neckel 1981;Wehrli 1985). The light levels on the sea surface correspond to solar noon at summer solstice and for a cloud-free day. Excluding the effects of cloud, little variation in E 0 (λ) will occur due to atmospheric effects in the PAR range (400-700 nm) because most of the absorption and scattering of photons by water vapor, ozone and oxygen in the atmosphere occur at longer or shorter wavelengths and are approximately spectrally neutral in the PAR range (Kirk 1994). We considered some atmosphericinduced variability (10 random atmospheric conditions) in sea surface light to provide a measure of variability in surface irradiance spectra (mean = 2071 μmol photon m À2 s À1 , SD = 149 μmol photon m À2 s À1 ). These conditions included: atmospheric pressure (990-1040 mbar, uniform distribution), precipitable water column vapor (1-5 cm, uniform distribution), aerosol optical depth (0.02-0.5, dimensionless, log-uniform distribution), relative humidity (50-100, dimensionless, uniform distribution), ozone (200-500 Dobson Unit, uniform distribution), ground albedo (0.01-0.08, dimensionless, uniform distribution), Angstrom exponent (À0.2 to 1.2, uniform distribution). The atmospheric model was described by Pinkerton and Hayward (2021) based on the model of Gregg and Carder (1990).

Modeling of benthic light spectra
Modeling was used to generate a large set of N (= 1 million) simulations of light spectra on the seabed using radiative transfer theory models and values of in-water light-absorbing components over wide ranges (Table 1; Supporting Information Figs. S1, S2) (Kirk 1994;Babin et al. 2003a,b;Lee 2005;Pinkerton et al. 2006). The downwelling irradiance reaching the seabed (E bed λ ½ ) at depth z and for the wavelength λ was modeled following the Beer-Lambert law (Eq. 1).
With E 0 λ ð Þ being the spectral instant solar irradiance just below the sea surface as described above; and K d λ ð Þ being the spectral diffuse attenuation coefficient for downwelling irradiance (m À1 ), which depends on the relative contribution of OACs characterized as seawater, CDOM, NAP, and Chl a. Note that Eq. 1 applies per wavelength, which allowed us to simulate efficiently spectral changes in attenuation and irradiance.
We related K d λ ð Þ to the total absorption (a) and backscattering (b b ) coefficients as Eq. 2 (Lee 2005 With a λ ð Þ (m À1 ) being the sum of the absorption by water molecules (a w ), phytoplankton (a ph ), detrital material (a d ), and gilvin (a g ) as well as b b λ ð Þ (m À1 ) being the sum of backscattering due to water molecules (b bw ), phytoplankton (b bph ), and detrital matter (b bd ), following Eqs. 3 and 4 (Kirk 1994): The absorption and backscattering spectra due to pure water were compiled using Pope and Fry's (1997) values. Next, N-simulated values of Chl a (chl, mg m À3 ) were randomly generated (Table 1). The absorption spectra due to phytoplankton were determined using spectral coefficients A(λ) and B(λ) from Bricaud et al. (1995) (Eq. 5) and a random perturbation factor (R NOMAD ) based on variations in the Chl a-specific absorption reported in the NASA bio-Optical Marine Algorithm Dataset (NOMAD version 1.3, Werdell and Bailey (2005)).
The absorption due to detrital material was modeled using Eq. 6 (Pinkerton et al. 2006) and based on N values of total suspended matter (tsm, g m À3 ; Table 1), non-algal particle mass-specific absorption coefficient (a Ã nap , m 2 g À1 ; Table 1) and the exponential slope absorption by detrital matter (S d , nm À1 ; Table 1).
The absorption due to CDOM, was modeled using Eq. 7 (Kirk 1994) and calculated for N random values of CDOM absorption at 440 nm (ag 440 , m À1 ; Table 1) and the CDOM exponential slope absorption (S g , nm À1 ; Table 1).
Backscattering by phytoplankton (b bph ) was modeled following Eq. 8 using the chlorophyll-specific backscattering due to phytoplankton (b Ã bphy , m 2 mg À1 ; Table 1) and the backscattering due to phytoplankton exponent (γ bphy , dimensionless; Table 1 Backscattering due to detrital matter (b bd ) was modeled following Eq. 9 using the mass-specific back-scattering due to nonalgal particles (b Ã bphy , m 2 g À1 ; Table 1) and the back-scattering due to non-algal particles exponent (γ nap , dimensionless; Table 1).
OACs such as chl, tsm, and CDOM can covary in the coastal zone as, for example, sediment discharge can carry nutrients which in turn stimulates phytoplankton growth. For simplicity, we omitted such co-variation, as the relative contribution of OACs to total attenuation can greatly vary in time and space (Devlin et al. 2009;Tilstone et al. 2012). A set of N depth distributions (z, m) was resampled randomly from the rocky reef locations around NZ using the NIWA bathymetry chart at 250 m resolution (Mitchell et al. 2012).
Absorption, action spectra, and photosynthetic metrics of macroalgal taxa The absorption spectrum of macroalgae depends on the morphology, biomass, and pigment content of the algal thallus as well as the optical method. To take into account the variability within taxon and among individuals, absorption spectra for different species within the Chlorophyta (27 species, n = 78 observations), Phaeophyta (31 species, n = 52 observations), and Rhodophyta (32 species, n = 44 observations) taxa were compiled from Enríquez et al. (1994). Standardized absorbances (median and 95% confidence intervals [CIs]) were converted into absorbance (optical density [OD]) by finding a vertical offset based on the linear relation between absorbance at 440 and 675 nm (Enríquez et al. 1994). Absorbance spectra in terms of absorbance (optical density) were then converted in absorptance, or the fraction of light absorbed by pigments, using Eq. 10 (Kirk 1994).
Synthetic and random absorptance spectra (N) were generated for each macroalgal taxon (A taxon ) in between their respective 95% CIs. The spectrally integrated fraction of light absorbed and used in photosynthesis by taxa (Ebed action,taxon , in μmol photon m À2 s À1 ) on the seabed corresponded to the metric of photosynthetic performance as being the amount of light in the PAR range absorbed and used in photosynthesis by a particular taxon (in μmol photon m À2 s À1 ). This was modeled using Eq. 11.
with R(λ) being the ratio between absorptance and action spectra (relative oxygen production) compiled from Kirk (1994) and for the three main macroalgal taxa. Here, A int,taxon represents the taxon-specific integrated absorptance spectrum and was used to normalize for variations in the total (integrated) absorptance between simulations in order to focus on variations in the spectral shape of absorption rather than absolute values. A second step of normalization was needed to compare photosynthetic performances of the three macroalgal taxa based on their spectral differences only: we applied a linear scaling to the photosynthetic performances of the green and red taxa based on their relationships to the brown taxon photosynthetic performance (Eq. 12; Supporting Information Fig. S2).
Ebed action,red,scaled ¼ 10 a Â Ebed action,red b Ebed action,green,scaled ¼ 10 c Â Ebed action,green d The photosynthetic efficiency of a particular taxon was calculated as the ratio between Ebed action,taxon over the total amount of light in the PAR range for the specific depth and attenuation condition Ebed (in %) and using the scaled red and green performances. The photosynthetic competitive efficiency was defined as the ratio between Ebed action,taxon of a particular taxon over Ebed action,taxon of another taxon (dimensionless) and using the scaled red and green performances. Finally, the photosynthetic relative performance ("Photosynthetic dominance") corresponded to the highest Ebed action,taxon between brown, red (scaled), and green (scaled) for each simulated condition. We showed photosynthetic metrics in the K d (PAR) vs. depth domain to help elucidate broad patterns, as depth and attenuation are the major factors affecting seabed light intensity. We focused on simulated conditions with at least 1 μmol photon m À2 s À1 at the seabed as a credible (but somewhat arbitrary) value for the minimum instant light requirement for macroalgae to be productive (light compensation point, I c ). We recognize that this lower irradiance threshold for macroalgal viability is taxon-and time-dependent (Markager and Sand-Jensen 1992).

Data transformation of inherent optical properties and ternary plots
A single value of K d (PAR) can be produced by any number of combinations of Chl a, or CDOM and NAP. We projected the data into the K d (PAR)-vs.-depth domain and then focused on subsets of this domain to look at second-order effects of light quality on the macroalgal metrics. Results were displayed using ternary plots as follows. We ranked and transformed the inherent optical properties (IOPs, a det , b bp , and a ph ) using Eq. 14 to compare the relative influence of IOP values on the total attenuation, taking out their difference in absolute values.
With p(r) being the transformed variable with values spanning between 0 (for the rank value = 1) and 1 (rank value = N). This transformation was used to spread the simulated data approximately evenly over the ternary plot domain to reveal patterns in the metrics across the range of drivers of attenuation. We used the R package "ggtern" version 3.3.5 (Hamilton 2016) to plot the transformed variables of IOPs for selected values of depth and attenuation coefficient in the PAR range.
In the ternary plots, we note that the IOPs (a det , b bp , and a ph ) are taken at 443, 555, and 488 nm (respectively) as these match the wavelengths commonly used in matching satellite products (Lee et al. 2002). In our analysis as in satellite products, a det includes CDOM and NAP absorption (i.e., is total detrital absorption) and b bp includes phytoplankton and NAP backscatter (i.e., is total particulate backscatter).

Results
Optically active constituents greatly absorbed and scattered light in the blue light region giving high attenuation between about 400 and 500 nm (Fig. 1). The increase of spectral attenuation at a constant depth (10 m) not only reduced the spectral amount of light at all wavelengths, but also shifted the benthic light spectrum toward yellow wavelengths (500-600 nm; Fig. 1). Typical simulated action spectra of green and brown macroalgae showed peaks in blue (400-500 nm) and red (600-700 nm) light whereas the action spectra of red macroalgae peaked in the 500-600 nm spectral region.
The amount of light in the PAR range reaching the seabed strongly depended on the nonlinear relationship between attenuation of light, K d (PAR), and depth ( Fig. 2A). A relatively low K d (PAR) value of 0.15 m À1 allowed light levels of 1 μmol photon m À2 s À1 to reach 50 m deep whereas increasing K d (PAR) led to a shallowing of the 1 μmol photon m À2 s À1 light level to $ 25 m (K d [PAR] = 0.3 m À1 ), $ 15 m (K d [PAR] = 0.5 m À1 ) and $ 7 m (K d [PAR] = 1 m À1 ) ( Fig. 2A). The photosynthetic performances of brown, red, and green algal taxa followed the same pattern of reduced performance for higher attenuation and depth (Fig. 2B-D). Given the same attenuation value, the 1 μmol photon m À2 s À1 light level is reached at shallower depths than for the total light in the PAR range (10 m vs. 15 m at $ 0.5 m À1 ). In fact, not all wavelengths in the PAR range were absorbed and used via the action spectra (Fig. 1).
Contrasting photosynthetic efficiencies at harvesting spectral light for a given attenuation and depth combination were found (Fig. 3, top row). In fact, the brown taxon photosynthetic efficiency was the highest ($ 18%) for relatively low K d (PAR) values (< 0.2 m À1 ) and quickly diminished with depth, especially for higher attenuation values (> 0.2 m À1 ), roughly from 16% at 3 m to 12% at 15 m for K d (PAR) $ 0.4 m À1 (Fig. 3, top left). A similar pattern of diminishing photosynthetic efficiency with depth at higher attenuation was found for the green taxon, with the difference that green efficiency stayed relatively high ($ 17%) at higher attenuation (> 0.2 m À1 ) and for shallow depths (< 5 m) (Fig. 3, top right). In contrast, the photosynthetic efficiency of the red taxon was lower at shallower ($ 12%, <5 m) than deeper depths ($ 15%-16%), especially at low to intermediate attenuation values (< 0.4 m À1 ) (Fig. 3, middle top row).
For a given PAR attenuation of $ 0.2 m À1 at 10 m deep, brown and green algae showed greater efficiencies ($ 18%) when total attenuation was dominated by particulates (b bp ) rather than phytoplankton (a ph ) or detrital material (a det ) (16%; Fig. 3A). This pattern was consistent at increased attenuation (0.4 m À1 ) at the same depth (Fig. 3B) as well as for the same attenuation and higher attenuation but deeper (20 m, Fig. 3C,D). The finer-scale impact of IOPs-at given PAR attenuation and depth conditions-in the photosynthetic efficiency of the red algae was less clear cut. For relatively low attenuation levels (0.2 m À1 ) at 10 and 20 m, the red efficiency was slightly higher ($ 16% vs. 15%) when attenuation was dominated by absorption by phytoplankton and detrital matter rather than by particulate scattering (Fig. 3A-C). In contrast, the pattern changed to higher efficiencies (16% vs. 14%) when attenuation was dominated by particulate scattering for relatively higher attenuation levels (0.4 m À1 ) at 10 and 20 m.
Photosynthetic competitive performance of the brown vs. red, green vs. red, and green vs. brown taxa showed clear patterns in the attenuation-depth space (Fig. 4, top row). Brown and green vs. red competitive performance ratios were > 1 at higher attenuation (> 0.2 m À1 ) for shallower depths (< 5 m; Fig. 4, top row) as well as for lower attenuation (< 0.2 m À1 ) deeper (up to 40 m), meaning that these conditions tended to advantage brown and green taxa compared to red taxon in terms of light harvesting compared to the average over the whole modeled depth-attenuation domain. Alternatively, the red taxon performed relatively better for higher attenuation (> 0.2 m À1 ) in deeper waters (> 5 m). Photosynthetic competitive performances of the green and brown taxa were higher at higher attenuation (> 0.2 m À1 ) in shallow waters (10-15 m; Fig. 4, top row), conditions which advantaged the green compared to the brown taxa in harvesting spectral light. Relatively lower attenuation (< 0.3 m À1 ) and deeper depths (> 15 m) favored brown algae as illustrated by lower ratios. At finer scales and for given attenuation and depth values, total attenuation dominated by particulate matter benefited the brown over the red taxa at 10 and 20 m as well as for attenuation values of 0.2 and 0.4 m À1 (Fig. 4A-D, left column). Similar patterns between the green vs. red taxa occurred as green prevalence (ratio above 1) is observed when total attenuation was dominated by particulate matter (b bp ) and red prevalence (ratio below 1) when attenuation was dominated by phytoplankton and detrital matter ( Fig. 4A-D, middle column). Fine-scale analysis of the competition between green vs. brown photosynthetic competitive performance was less clear cut at lower attenuation and depths 10-20 m (Fig. 4A,C, right column) but patterns emerged at higher attenuation (Fig. 4B,D, right column) with conditions vs. depth space. Data points with Ebed(PAR) and photosynthetic performances less than 1 μmol photon m À2 s À1 are shown in gray.
favoring green algae (ratio above 1) when attenuation was dominated by phytoplankton and detrital matter, and favoring brown algae (ratio below 1) when attenuation was dominated by particulate matter.
Overall, the highest relative performance (photosynthetic dominance) was achieved by the brown algae for lower attenuation (< 0.2 m À1 ) and deeper areas (< 10 m) (Fig. 5, top) whereas the green algae showed the highest relative photosynthetic performance for shallower depths (< 10 m) and all attenuation values. Red algae had the highest relative photosynthetic performance under deeper (> 10 m) and higher attenuation (K d [PAR] > 0.2 m À1 ) conditions (Fig. 5, top). Finer-scale analysis indicated that brown algae performed better than the red algae when there was relatively lower attenuation (0.18 m À1 ) dominated by particulate matter deeper (20 m) (Fig. 5A). In shallow waters (10 m) with low attenuation (0.22 m À1 ), red algae had higher relative performance compared to green and brown algae where attenuation was dominated by phytoplankton and detrital matter (Fig. 5B). In these environments, there were no clear differences in photosynthetic relative performance between the brown and green taxa (Fig. 5B). Finally, in higher attenuation (0.4 m À1 ) and relatively shallow waters (8 m), there were no clear differences in relative photosynthetic performance between red and green taxa in respect to the type of taxa as a function of attenuation (K d [PAR]) and depth (top row) as well as of a ph , a det , and b bp (in ternary plots) and for selected depth and K d (PAR) intervals (A-D, see Fig. 3 for values). The photosynthetic competitive performance is the ratio between photosynthetic performances for two specific taxa (first-named divided by second-named taxon). Scaled photosynthetic performances are used (red and green to brown).
IOPs dominating attenuation. The brown algae seemed to be favored when attenuation was dominated by particulate matter (Fig. 5C).

Discussion
Variability in the composition of optically active constituents in the water column changed the spectral shape of light at the seabed, with significant implications for the relative photosynthetic efficiency of macroalgal groups. We showed that red algae have enhanced photosynthetic efficiency relative to both green and brown algae under most light-limited scenarios (confirming Hypothesis 1), particularly at high overall attenuation coefficients and at greater depths. Furthermore, we showed that high concentrations of Chl a in the overlying water column (aph, e.g., phytoplankton) can favor red algae over brown and green algae (confirming Hypothesis 2). The overlapping light absorption by water column phytoplankton and both green and brown algae action spectra has implications for light harvesting at the seabed and suggests that eutrophication-induced phytoplankton bloom could substantially affect the structure and functioning of benthic macroalgal assemblages, especially at the extremities of light limitation in coastal scenarios (confirming Hypothesis 3). Changes in light availability in terms of quantity, which is usually integrated over the PAR range (400-700 nm), is well known to significantly affect benthic macroalgae assemblages (Airoldi 2003;Clark et al. 2013), resilience (Desmond et al. 2015;Tait 2019), and their associated ecosystem services like carbon cycles (Blain et al. 2021;Smith et al. 2021;Weigel and Pfister 2021). However, benthic light also has a spectral quality component, and it is still unclear how changes in spectral light quality tied to changes in OACs in the water column are related to patterns of occurrence in macroalgal communities. Here, we highlight how changes in light quality separated from quantity (PAR irradiance) can benefit different macroalgal groups across various benthic scenarios of OACs and depths. We found that brown macroalgae were more efficient at harvesting light at lower total attenuation values (< 0.2 m À1 ) and in deeper conditions (> 10 m), whereas green macroalgae were favored at shallower depths (< 10 m) and for low to high attenuation values. Finally, red macroalgal photosynthetic performance was favored at relatively higher attenuation values (> 0.2 m À1 ) and deeper depths (> 10 m).
The patterns in relative photosynthetic performance illustrated here are explained by the greater attenuation of blue and green light by OACs, which-when exacerbated by deeper depths-compressed the benthic light spectrum around the yellow light band (500-600 nm), corresponding to the peak in the action spectrum of red macroalgae. The shift in benthic spectral light induced by greater attenuation of blue and green wavelengths can explain the depression or disappearance of canopy-forming brown algae in reefs with low water clarity, especially around urbanized and modified terrestrial catchments, or those affected by dredging activities (Lyngby and Mortensen 1996;Airoldi 2003;Connell et al. 2008;Desmond et al. 2015;Tait 2019). Changes in spectral light quality can also affect the early life stages of brown algae (gametophytes) due to requirements for blue-wavelengths (Lüning and Dring 1975;Schiel and Foster 2015). Despite a general erosion of the CCA hypothesis as a guiding principle for the organization of marine macroalgae (Raven and Hurd 2012), the variable attenuation of the PAR spectrum can potentially play an important role in the relative composition of macroalgal assemblages, particularly under scenarios where light is limiting-that is, when irradiance is not sufficient to saturate the algal photosynthetic apparatus.
Diverse photosynthetic pigment profiles enhance complementary light-use among mixed species assemblages, including among layered macroalgal assemblages  and phytoplankton assemblages (Stomp et al. 2004;Stomp et al. 2007). Moreover, measurements of the spectrum of light available to subcanopy algae-which is dominated by the wavelengths of light not absorbed by overlying canopies of brown algae (Tait et al. 2017)-results in the dominance of red algae within these subcanopy environments (Tait and Schiel 2018).
Variation of light quality with depth may also lead to ecological shifts in freshwater and marine phytoplankton communities (Luimstra et al. 2020;Stockenreiter et al. 2021), and water column optical properties have been shown to greatly vary in coral reef waters (Russell et al. 2019, Hochberg et al. 2020. The role of light quality on coral-algal symbionts is suggested to be important, but often overlooked, in coral reef ecology (L opez-Londoño et al. 2021). By bridging optical modeling and macroalgal physiology, we showed striking differences in photosynthetic efficiency (Fig. 3), competitive performance (Fig. 4) and relative "dominance" (Fig. 5) of the three main macroalgal taxa.
Overall, our results provide some support for the CCA hypothesis and the work of Levring (1968), as they confirm that algal pigmentation matters under a subset of scenarios, particularly at the limits of light availability in coastal waters. However, changes in light quality are very often coupled with changes in overall light quantity so that the importance of light quality is small compared to light quantity when light saturates photosynthesis, as argued by Dring (1981), and remains one of the strongest arguments against the CCA. Although we note the ongoing relevance of Dring (1981), we showed that there is relevancy of the CCA at the extreme attenuation scenario, for waters mostly dominated by one type of OAC. Questions remain about the frequency at which such optical conditions are observed, in fact the dynamics of OACs is complex, greatly varying in space and time (Devlin et al. 2009;Tilstone et al. 2012;Gall et al. 2022; Supporting Information Fig. S4).
As well as the implications of our research to the CCA debate, this study further supports the need to develop satellitederived products that not only estimate the light quantity on the seabed (Gattuso et al. 2020), but also deconvolute its spectral components by yielding information about the light quality on the benthos. Ocean color remote sensing paired with semi-analytical algorithms better allow the retrieval of IOPs in "optically complex" Case 2 waters (Werdell et al. 2018;Najah and Al-Shehhi 2022), and subsequent satellite-derived products of OACs can be used for spatiotemporal assessment of the relative contribution of Chl a, NAP, and CDOM to light attenuation. Understanding OAC variability in the coastal zone is an Earth-observation research priority (Malthus and Mumby 2010) and powerful trend analysis could be investigated on the longest ocean color time series (i.e., MODIS Aqua, Gall et al. (2022)). Along with remote sensing, optical modeling can help in investigating the relationships between IOPs and spectral benthic irradiance as it allows modeling wide but realistic optical properties. Radiative transfer modeling also extends the capability of in situ measurement of spectral light and IOPs because bio-optical field measurements, despite being essential, are labor-and time-intensive. Our approach could be extended in the future using models of instantaneous seabed light for different depth/attenuation scenarios to drive simulations of primary productivity by macroalgae, notably through the use of seasonally and depth-dependent productivity-irradiance curves (Blain and Shears 2019) as well as more deeply investigating the implications of different daily light dose delivery for macroalgal-productivity (Desmond et al. 2017). Such insights will be likely to help manage anthropogenic impacts of reduced light availability (e.g., those induced by dredging) and by assessing the relevancy of environmental windows, such as avoiding anthropogenic perturbations during macroalgal lifestage sensitive periods (Airoldi 2003;Fraser et al. 2017).
Although we showed clear scenarios and mechanisms for shifting macroalgal composition based on water column constituents and depth gradients, limitations, and uncertainties remain. Our study did not consider temporal variability of light delivery to the seabed which is notably dynamic, associated with diurnal cycles, seasonality, tides, and turbidity events (e.g., sediment discharge, phytoplankton blooms; Anthony et al. 2004). Likewise, our study did not integrate the photoadaptive capability of thalli or broad life-history strategies which are critical to realized ecological patterns (Raven and Hurd 2012). Our study has integrated variation in OACs over space for a snapshot in time (solar noon, summer) across known rocky reef habitat in NZ (Supporting Information Fig. S4; Gall et al. (2022)) but has not accounted for annual variability in both light quantity and quality. For a given attenuation coefficient and depth (similar light quantity), we found that photosynthetic performance varied by just a few points ($ 13%-17%; Fig. 3) with the largest differences caused by attenuation mostly dominated by one type of OAC. Further work is needed to determine if substantial differences of photosynthetic performances between taxa are predicted for longer period of times. However, our approach has allowed for quality and quantity of light to be assessed independently (an important prerequisite identified by Dring (1981)) and provides a theoretical assessment of photosynthetic efficiency across realistic scenarios of water column constituents across water bodies that are known to be affected by sediments and eutrophication (Schiel and Howard-Williams 2016;Safi et al. 2022). Experimental work will be essential to validate our theoretical observations by simulating different spectral light regimes and measuring productivity of different species and taxa under contrasting attenuation scenarios, for example, by using spectrally adjustable lights in the laboratory.
We conclude that phytoplankton and sediments not only reduce the amount of light on the seabed but also remove crucial colors, affecting macroalgal taxa differently, with the potential to cause ecological shifts. Loss of brown canopyforming seaweeds is of particular concern globally (Krumhansl et al. 2016;Filbee-Dexter and Wernberg 2018) with major implications to ecological function and carbon cycling (Duarte et al. 2022;Pessarrodona et al. 2022). Improving our understanding of a shifting underwater light environment is needed in the face of climate change and increasing anthropogenic pressures in the coastal zone. Restoring healthy levels of coastal water quality will improve benthic light quantity and quality, which in turn might play a crucial role in buffering the adverse effects of increasing coastal marine heatwaves, especially for kelps and fucoids (Thomsen et al. 2019;Tait et al. 2021). The approach used here paves the way to apply remote sensed water column OACs, bathymetric data, and physical habitat information to pinpoint habitats and depths sitting at thresholds of competitive performance. This information would enable practical tests of the possible impacts of degradation of the light environment on macroalgal communities at a global scale.