Optical Properties of Corals Distort Variable Chlorophyll Fluorescence Measurements

Variable chlorophyll fluorescence is distorted by the optical properties of corals as demonstrated by experimental studies on hydrogel slabs and optical simulations.

Abstract

Pulse-amplitude–modulated (PAM) fluorimetry is widely used in photobiological studies of corals, as it rapidly provides numerous photosynthetic parameters to assess coral ecophysiology. Coral optics studies have revealed the presence of light gradients in corals, which are strongly affected by light scattering in coral tissue and skeleton. We investigated whether coral optics affects variable chlorophyll (Chl) fluorescence measurements and derived photosynthetic parameters by developing planar hydrogel slabs with immobilized microalgae and with bulk optical properties similar to those of different types of corals. Our results show that PAM-based measurements of photosynthetic parameters differed substantially between hydrogels with different degrees of light scattering but identical microalgal density, yielding deviations in apparent maximal electron transport rates by a factor of 2. Furthermore, system settings such as the measuring light intensity affected F₀, F_m, and F_v/F_m in hydrogels with identical light absorption but different degrees of light scattering. Likewise, differences in microalgal density affected variable Chl fluorescence parameters, where higher algal densities led to greater F_v/F_m values and relative electron transport rates. These results have important implications for the use of variable Chl fluorimetry in ecophysiological studies of coral stress and photosynthesis, as well as other optically dense systems such as plant tissue and biofilms.

The ecological success of coral reefs is largely due to the successful symbiotic relationship between the coral animal host and its photosymbiotic microalgae belonging to the genus Symbiodinium. This highly efficient symbiotic interaction is susceptible to changes in environmental conditions, such as excess solar radiation and above-average seawater temperatures, which can lead to the breakdown of the coral–algal symbiosis and the visible paling of the coral colony known as “coral bleaching” (Weis, 2008). Given the importance of Symbiodinium photosynthesis for coral health, coral photosynthesis has been studied intensively from molecular to environmental scales (Falkowski et al., 1990; Dubinsky and Falkowski, 2011). Coral photosynthesis can be studied with techniques quantifying photosynthetic O₂ production or carbon fixation (Kühl et al., 1995; Hoogenboom et al., 2012; Osinga et al., 2012), but photophysiological measurements based on variable chlorophyll (Chl) a fluorescence are now widely used in coral research and many other areas of terrestrial and aquatic photosynthesis research (Warner et al., 1996, 2010; Ralph and Gademann, 2005; Szabó et al., 2014). In contrast to gas exchange or C-fixation measurements that require significant sample handling, variable Chl fluorescence relies on optical light pulsing schemes that are applied externally with minimal sample manipulation or directly in the natural habitat, and a variety of commercial instruments for cuvette-based, fiber-optic or imaging measurements are available (e.g. Schreiber, 2004). In coral reef science, pulse-amplitude–modulated (PAM) Chl a fluorimeters are by far the most commonly used instrument to probe photosynthesis (Warner et al., 2010).

Variable Chl fluorimetry quantifies the fate of absorbed light energy trapped by the photosynthetic apparatus via changes in Chl a fluorescence, which tracks the redox status of PSII and the balance between photochemical and nonphotochemical quenching processes. PAM-based measurements employ the so-called “saturation pulse” method (Schreiber, 2004). The PAM technique generates multiple photochemical charge separations (multiple turnover) and fully reduces Q_A via the application of 50–1,000-ms multiple turnover flashlets (“fat flashes”; see Kromkamp and Forster, 2003). The fluorescence yield before the saturation pulse indicates the level of fluorescence when Q_A is maximally oxidized and PSII reaction centers are fully open. Such minimum fluorescence yield is denoted as F₀ and F, referring to dark- and light-acclimated samples, respectively. The saturation pulse is assumed to lead to the complete closure of PSII reaction centers, such that photochemical quenching is fully inhibited (Schreiber et al., 1993). Consequently, fluorescence emission is maximal and this parameter is known as the “maximal fluorescence yield,” where F_m and F_m′ refer to dark- and light-acclimated samples, respectively (Schreiber, 2004). From these measurements, many derived parameters describe the photophysiology of the investigated sample.

In coral reef science, the most frequently used fluorescence parameter is the maximum PSII quantum yield, which is calculated from saturation pulse measurements on dark-acclimated samples as follows: F_v/F_m = (F_m-F₀)/F_m. The F_v/F_m parameter is considered a key proxy for coral health, and differences in F_v/F_m between coral measurements are interpreted as a change in coral fitness (Jones et al., 2000; Wiedenmann et al., 2013). The effective quantum yield of PSII, Φ_PSII = (F_m′-F)/F_m′, is determined via a saturation pulse measurement under a known actinic irradiance of photosynthetic active radiation (PAR; Genty et al., 1989). An estimate of the relative PSII-derived photosynthetic electron transport rate is calculated as r_ETR = Φ_PSII × PAR (Ralph and Gademann, 2005), whereas determination of the absolute ETR requires information about the PSII absorption cross section (Szabó et al., 2014). When calculated over a range of actinic irradiance levels, r_ETR versus irradiance curves (i.e. light curves) can be determined, which enables the calculation of the maximum electron transport rate (ETR_max) and the light-use efficiency factor (α; i.e. the initial slope of the rETR-versus-irradiance curve). Measurement protocols for the application of PAM on corals are well-described (e.g. Warner et al., 2010), and the maximum PSII quantum yield and r_ETR-versus-irradiance–curve parameters are frequently used to interpret the health and photo-physiological acclimation state of Symbiodinium within the coral host (Ralph et al., 2005).

However, the application of variable Chl fluorescence is based on the assumptions that all photosynthetic entities (cells/chloroplasts) are (1) equally exposed to the incident actinic light levels, (2) equally exposed to the measuring light (ML) and emitting fluorescence equally, and (3) effectively saturated by the saturation pulse (Schreiber et al., 1996; Serôdio, 2004). In other words, it is assumed that (1) each photosynthetic cell has identical fluorescence excitation/emission, and (2) the generated fluorescence from each cell has equal probability to be detected by the fluorimeter. These assumptions are only met, if the light distribution within the sample is homogenous, such as in optically dilute algal cultures (e.g. Ting and Owens, 1992). In contrast, most photosynthetic tissues exhibit strong scattering and absorption, leading to a heterogenous distribution of irradiance within the sample (e.g. Serôdio, 2004; Evans, 2009; Oguchi et al., 2011; Szabó et al., 2014; Evans et al., 2017; Lichtenberg et al., 2017). For instance, steep light gradients exist in biofilms (Kühl and Jørgensen, 1994), which can lead to an effective overestimation of Ф_PSII and ETR (Serôdio, 2004). Such gradients can also affect measurements of the Chl a fluorescence kinetics (Sušila et al., 2004). Reabsorption of fluorescence emission can pose a challenge for Chl a fluorimetry in optically dense tissues (Naus et al., 1994; Bartošková et al., 1999). A method for correcting variable fluorescence measurements in optically dense algal media under constant optical geometries (e.g. cuvettes) has been proposed (Klughammer and Schreiber, 2015), but this approach assumes a simple exponential light attenuation (Lambert–Beer’s law), which is too simplistic for light-scattering photosynthetic tissues (Kühl and Jørgensen, 1994; Wangpraseurt et al., 2016a).

Knowledge of the tissue inherent optical properties (IOPs) allows us to predict light propagation by solving the radiative transfer equation. The IOPs are defined as the probability of light absorption per infinitesimal path length (μ_a, [mm⁻¹]); the probability of light scattering per infinitesimal path length (μ_s, [mm⁻¹]); and the anisotropy of scattering (g), i.e. the average cosine 〈cos θ〉 of the scattering angle (θ), and the refractive index (n). In strongly light-scattering samples, μ_s is combined with g to define the reduced scattering coefficient μ_s′ = μ_s · (1−g; Jacques, 2013). The reduced scattering coefficient describes photon diffusion in a random walk of step sizes of 1/μ_s′, where each step involves isotropic scattering. Recent progress in coral optics (Wangpraseurt et al., 2014b, 2016a; Swain et al., 2016) revealed that coral tissues and skeletons can be strongly light-scattering and that μ_s′ is variable between coral species. For instance, μ_s′ of coral skeletons can vary by one order of magnitude (Marcelino et al., 2013), thus substantially affecting the amount of light that is backscattered into the overlying algal layer (Marcelino et al., 2013; Wangpraseurt et al., 2016a). Thick-tissued corals can have light-scattering host pigments (e.g. green fluorescent protein [GFP]) situated on top of the algal layer (Lyndby et al., 2016; Wangpraseurt et al., 2017b). In such a scenario, light is effectively scattered before it reaches Symbiodinium cells, leading to a strong surface enhancement of scalar irradiance, E₀ (= fluence rate; Lyndby et al., 2016).

The absorption coefficient (μ_a) of corals is largely dependent on Symbiodinium cell density and Chl a content per cell (Wangpraseurt et al., 2016a). Algal densities in corals vary seasonally (Chen et al., 2005) and in response to environmental stress (Glynn, 1996). Symbiodinium density affects vertical light attenuation, where densely pigmented corals are characterized by steep light gradients, while less-pigmented corals have a diffusely enhanced tissue light environment (Wangpraseurt et al., 2017a). The photosynthetic yield of Symbiodinium can be vertically stratified within the coral tissue (Lichtenberg et al., 2016; Wangpraseurt et al., 2016b). During coral bleaching, microalgal symbionts experience photoinhibition (Warner et al., 1999) and such damage is likely more prominent in the light-exposed top layers of coral tissue (Lichtenberg et al., 2016; Wangpraseurt et al., 2016b). Corals thus represent a challenging study organism for variable Chl fluorimetry; yet, to our knowledge, no studies have aimed at ground-truthing the central assumptions of PAM measurements on corals.

The use of natural coral samples for such study would be easily contrived by variability of the IOPs of individual samples (e.g. changes in µ_a, µ_s′). To avoid this variability, we developed optical analogs to corals using a biomedical tissue optics approach. Optical phantoms are often created to solve problems related to the propagation of light in scattering tissues, e.g. for calculating the light dose in photodynamic therapy and cancer treatment (Pogue and Patterson, 2006). The optical response of human tissue is mimicked by optical phantoms consisting of (1) a gel-like planar matrix (e.g. gelatin, agar, or agarose), (2) light-scattering particles (e.g. SiO₂, TiO₂, polystyrene microspheres), and (3) light absorbers (e.g. intralipid, India ink; Tuchin, 2007).

In this study, we created multiple planar hydrogel slabs to replicate the bulk optical properties of corals and characterized their variable Chl fluorescence-derived parameters, including F_v/F_m, Y(II), and r_ETR (Supplemental Table S1). Specifically, we examined the role of light scattering by replicating three major coral categories: (1) corals with strongly backscattering skeletons (“skeleton” hydrogel), (2) corals with low-scattering skeletons (“transparent” hydrogel), and (3) corals with light scattering in skeleton and tissue (e.g. due to GFP host pigments; “GFP” hydrogel). We also assessed the role of light absorption for Chl a fluorimetry by creating hydrogels with different microalgal densities and PSII efficiencies. Furthermore, we developed a light-propagation model (chlorophyll fluorescence Monte Carlo [Chf-MC] simulation) that allows for prediction of the generated fluorescence and detected fluorescence as a function of tissue optical properties. Although we focused on the widely applied PAM method, it is also relevant for other variable Chl fluorescence methods such as fast repetition rate fluorimetry (Gorbunov et al., 2001). The optical phantom approach (Fig. 1) and numerical models can easily be altered to address identical questions in other light-scattering samples such as leaves and biofilms.

Figure 1.

Coral tissue organization and artificial tissue design. A–C, Basic organization of Scleractinic corals. A, Small fragment of a faviid coral (scale bar = 1 mm). B, Close-up of the cross section, revealing the white coral skeleton, the brown algal layer on top of the skeleton, and the GFP like-pigment granules (“GFP”) on the coral tissue surface (scale bar = 1 mm). C, Close-up of GFP granules (scale bar = 200 μm). D–I, Coral-tissue–mimicking hydrogels. Schematics of three-layer “GFP” (D), Two-layer “skeleton” (E), and two-layer “transparent” (F) designs. G–I, Respective photographs of thin cross sections of hydrogels. J–M, Hydrogels for investigating the effect of changes in coral absorption. J and K, High microalgal-density–tissue design (J) and top-view photograph (K). L and M, Medium microalgal density hydrogel (L) and top-view photograph (M). N–P, Hydrogels for investigating the effect of coral stress. N, “Healthy” coral design with two layers of Rhodomonas sp., O, “Stressed” coral design with one layer of Nannochloropsis sp. on top of one layer of Rhodomonas sp., P, “Bleaching” design with one layer of Nannochloropsis sp. of a reduced cell density on top of R. salina.

RESULTS

Effects of ML Settings and Light Scattering on Variable Chl Fluorescence

Hydrogels with identical absorber density but different light-scattering properties showed up to 5-fold differences in F₀ for the same ML intensity (Fig. 2, A–D). Highest F₀ values were achieved for the skeleton hydrogel (F₀ = 0.162; Fig. 2, B and D). ML intensity also affected measurements of the maximal fluorescence yield (F_m) and the calculation of F_v/F_m (Fig. 2, E and F). For ML = 3–4, F_v/F_m values were ∼0.74 for all three coral-mimicking hydrogels, while for ML ≤ 3 and > 5, F_v/F_m values differed by up to 0.1 (Fig. 2F).

Figure 2.

Effect of light scattering on variable Chl fluorescence parameters of dark-acclimated coral-tissue–mimicking hydrogels. A–C, Example images of minimal fluorescence yields (F₀) for ML = 4, showing GFP (A), skeleton (B), and transparent (C) hydrogels. The white circle shows the area over which F₀ was integrated. D–F, Effect of measuring light intensity on F₀ (D); F_m, maximal fluorescence yield (E); and the maximum PSII quantum yield, F_v/F_m (F). Note that no measurements are shown at ML > 4 (= 0.8 μmol photons m⁻² s⁻¹) and ML > 8 (= 1.6 μmol photons m⁻² s⁻¹) for skeleton and transparent hydrogels, respectively, due to indications of actinic effects in these hydrogels at higher ML levels.

The in vivo light microenvironment measured with fiber-optic microsensors differed for the three coral-mimicking hydrogels, and photon E₀ (400–700 nm) at the hydrogel surface reached 109% (±0.85 se; n = 8) of the downwelling photon irradiance, E_d (Supplemental Table S1) for the transparent hydrogel, 142% E_d (±9 se; n = 8) for the skeleton hydrogel, and 244% E_d (±12.3 se; n = 8) for the GFP hydrogel (Fig. 3). The steepest light attenuation was measured in the GFP hydrogel, and the lowest, in the transparent hydrogel (Fig. 3). The top layer (0–750 µm) of the light-scattering GFP hydrogel (Fig. 1D) created a subsurface maximum in E₀ (at ∼250 µm below the hydrogel surface); this was followed by rapid light attenuation within the light-absorbing algal layer (750–1,500 µm). For the skeleton hydrogel, light attenuated to 125% E_d within the first 300 µm, after which light scattering by the underlying layer caused a subsurface maximum of ∼700 µm from the hydrogel surface that reached 170% E_d (Fig. 3).

Figure 3.

In vivo light microenvironment in coral-tissue–mimicking hydrogels with different scattering properties. Photon E₀ of PAR (400–700 nm) was normalized to the E_d of PAR and plotted against the vertical depth (µm) of the coral mimics. The algal layer is distributed between depth = 0–750 μm for the transparent (red) and skeleton (blue) mimic, while the algal layer is between 750 and 1,500 μm in the GFP (green) mimic. Four replicate gels were measured at two random spots (total n = 8 ± se).

Steady-state light curves revealed that the effective quantum yield of PSII (Ф_PSII) and the derived relative electron transport rates (r_ETR) differed between the three coral-mimicking hydrogels (Fig. 4), where the shape of the curves depended on the light field parameter used to quantify the actinic light level. When plotted as a function of E_d, the Ф_PSII was higher for the GFP hydrogel than for the skeleton and transparent hydrogel, and this difference was larger for r_ETR calculations. For instance, at E_d = 1,000 µmol photons m⁻² s⁻¹, r_ETR was ∼1.5 times higher for the GFP versus the other two hydrogels (Fig. 4C). Correction of ETR for in vivo E₀ led to similar patterns between the GFP and skeleton hydrogel, which now both showed higher r_ETR values than the transparent hydrogel (Fig. 4D).

Figure 4.

Effect of light scattering on the effective quantum yield Ф_PSII (A and B) and r_ETR (C and D) of coral-tissue–mimicking hydrogels. Measured Ф_PSII and calculated r_ETR were plotted as a function of the E_d (A and C) and with the corrected in vivo E₀ (B and D). Symbols with error bars represent means ± se (n = 5 biological replicates).

Effects of Light Absorption on Variable Chl Fluorescence

We constructed hydrogels with identical scattering properties but different light absorption properties (Fig. 1, J–M). Surface photon E₀ (400–700 nm) in the hydrogel with medium algal density (1.0 × 10⁶ cells cm⁻²) was ∼1.4-fold higher than that in the hydrogel with high algal density (3.5 × 10⁶ cells cm⁻²; 205% E_d ± 0.27 se versus 146% E_d ± 0.51 se; n = 4). This difference increased as a function of vertical depth, and measurements at depths of >2,000 µm showed up to 3-fold enhanced E₀ values in the medium- versus high-algal–density hydrogel (Fig. 5A).

Figure 5.

Effect of coral light absorption on the light microenvironment and variable Chl fluorescence measurements. Photon E₀ (400–700 nm) was normalized to the E_d and plotted against the vertical depth (μm) in the coral mimics. A–D, The algal layer is distributed between 0- and 750-μm depth (A). The maximum quantum yield of PSII, F_v/F_m (B). Relative electron transport rates (r_ETR) calculated as a function of the E_d (C) and corrected for in vivo E₀ (D). All measurements were performed in coral hydrogels mimicking high algal density (3.5 × 10⁶ cells cm⁻²) and medium algal density (1.0 × 10⁶ cells cm⁻²). Symbols represent means ± se (n = 4 biological replicates) in (A to C). The curve fit shown in (D) was the best fit to the experimental data (R² = 0.98 and 0.80) yielding values of ETR_max = 80 and 59, and α = 0.22 and 0.11 for the high- and medium-algal–density coral hydrogels, respectively.

Microalgal density had a significant effect on estimates of the maximum quantum yield, where hydrogels with 3.5 × 10⁶ cells cm⁻² showed ∼0.04 units higher F_v/F_m values than hydrogels with 1.0 × 10⁶ cells cm⁻²; Student’s t test: t(6) = 11.25, P < 0.01 (Fig. 5B). Likewise, microalgal density affected r_ETR, and at E_d = 500 µmol photons m⁻² s⁻¹, r_ETR was ∼2.7-fold higher for the high- versus medium-algal–density hydrogel (r_ETR = 41.1 ± 0.1 versus 14.7 ± 0.8; n = 4; Fig. 5C). Additionally, r_ETR values were corrected for the in vivo E₀, as determined with E₀ microsensors for each respective photic zone (i.e. the hydrogel layer that contained microalgal cells). Correction for in vivo E₀ slightly improved this discrepancy, but calculations of rETR_max and the light-use efficiency factor (α) were still ∼1.35 (rETR_max = 80 versus 59) and 2-fold higher (α = 0.22 and 0.11) for high- versus medium-algal–density hydrogels.

Effects of Bio-optical Properties and Simulated Coral Bleaching on Variable Chl Fluorescence

Hydrogels mimicking a stressed but not bleached coral, i.e. harboring a top layer with high algal density but low photosynthetic potential, showed low r_ETR and onset of photoinhibition (Fig. 6). However, hydrogels mimicking a partially bleached and stressed tissue, i.e. harboring a top layer with reduced microalgal density and photosynthetic potential, showed moderate r_ETR with rETR_max = 38 (at E_d > 500 µmol photons m⁻² s⁻¹) and no signs of photoinhibition (Fig. 6).

Figure 6.

Combined effects of bio-optical properties and simulated coral bleaching on measured r_ETR. Hydrogels mimicking healthy tissue contained two dense layers of R. salina (4 × 10⁶ cells cm⁻²); hydrogels mimicking stressed tissue contained one dense layer of Nannochloropsis oculata (2 × 10⁶ cells cm⁻²) and one dense layer of R. salina (2 × 10⁶ cells cm⁻²); and hydrogels mimicking bleached tissue contained one layer of N. oculata at low density (0.3 × 10⁶ cells cm⁻²) and one dense layer of R. salina. Data are means ± se (n = 2–3 hydrogel replicates).

DISCUSSION

Variable Chl fluorimetry is a key tool for probing photosynthesis in vivo (Baker, 2008). However, the assumptions underlying the calculation of variable Chl fluorescence parameters might not be fulfilled when measuring externally on highly stratified and dense photosynthetic tissues, such as corals, biofilms and plant tissues (Serôdio, 2004; Evans, 2009; Szabó et al., 2014). Our results showed that F₀, F_m, and Fv/F_m were affected by the scattering properties of coral-mimicking hydrogels (Fig. 2). In a first approximation, we can describe the detected F₀ signal by using three simple terms: (1) the ML intensity incident on an algal cell, i.e. fluorescence excitation; (2) the fluorescence emission per cell, which is governed by the biophysical properties of the cell, such as Chl a content and dark acclimation (Warner et al., 2010); and (3) the probability that such emitted fluorescence is detected by the imaging instrument (fluorescence escape; see Supplemental Text). Because algal cells can collect light from all directions, the excitation term is not described by the downwelling irradiance of the ML, but by the E₀ (= fluence rate) of ML (Kühl et al., 1995), which in turn is affected by the tissue optical properties (Jacques, 2013; see optical simulations in Supplemental Text). In the first experiment (Fig. 1), we kept μ_a constant while modulating μ_s′, which created characteristic differences in the light microenvironment between three different coral tissue mimics (Fig. 3). The enhancement of photon irradiance at the tissue surface of the skeleton hydrogel was due to the strong backscattering (Fig. 3). In contrast, light attenuation was described by a simple exponential attenuation for the transparent hydrogel (Fig. 3).

The observed differences in the F₀ signal between the skeleton and transparent hydrogels for a given ML intensity were due to two mechanisms (Fig. 7). Firstly, the fluence rate of the ML was enhanced in the absorbing layer (0–750-μm depth) for the skeleton versus the transparent hydrogel (Fig. 3). The higher fluence rate led to an increased chance of photon absorption and thus higher levels of fluorescence generation (see optical simulations in Supplemental Fig. S2). Secondly, although an individual algal cell acts as an isotropic point source (i.e. emitting fluorescence equally well in all directions; Schreiber, 2004), the detected fluorescence signal depends on the propagation of fluorescent light from this point source, through the tissue toward the fluorimeter (Welch et al., 1997). Because intact corals are typically monitored externally using a fiber or camera in backscattering configuration, the reflectivity of the skeleton controls the upwelling fluorescence toward the detector. For the transparent hydrogel, the downwelling fluorescence (F_d) was essentially lost, while backscattering by the skeleton hydrogel led to an effective redirection of the otherwise-lost F_d (Fig. 7).

Figure 7.

Propagation of PAM-based ML in biological tissues with different scattering coefficients. A and B, Excitation (blue) and Chl fluorescence (red) for transparent hydrogel (A) and skeleton hydrogel (B). The optical properties of the light absorbing top layer (layer 1) are constant (i.e. identical algal density and biophysical properties of an algal cell) for both gels, but layer 2 is either transparent or light-scattering. For the transparent hydrogel, ML absorption by an algal cell (s) is a function of the primary incident beam (solid blue line), while indirect light (dotted blue lines) is lost through the transparent layer. For the skeleton hydrogel, indirect light is redirected via backscattering by layer 2. Such scattering enhances the chance of ML absorption and thus leads to greater fluorescence emission (bold red arrows). Fluorescence emission is an isotropic process but the propagation of fluorescent light is affected by tissue optical properties. For the transparent hydrogel, only primary upwelling fluorescence (F_u1) contributes to the detected fluorescence signal, while for the skeleton hydrogel the F_d is redirected and adds to the upwelling fluorescence (F_u2). C, A steep light gradient (green line) leads to an underrepresentation of fluorescence detection from lower cell layers compared to an homogenous light environment (black line), given that the operational volume (dotted lines) from which fluorescence is collected is a function of the theoretical instrument detection limit (E_limit), which is modulated by the in vivo photon E₀.

For the GFP hydrogel, the light-scattering elements were placed on top of the light-absorbing algal layer (Fig. 1). Scattering diffuses the incident light, and diffuse light penetrates less into biological tissue than collimated light (Tuchin, 2007; Wangpraseurt and Kühl, 2014; Supplemental Fig. S2). Thus, although intense scattering in the top tissue layer would enhance the chance of fluorescence emission and subsequent upwelling of generated fluorescence, it also leads to a steep attenuation of the ML within the algal layer (Lyndby et al., 2016; Fig. 3; Supplemental Fig. S2). The vertical attenuation of E₀ (400–700 nm) within the light-absorbing layer was described according to Lambert–Beer’s law for the transparent and GFP hydrogels, yielding an attenuation coefficient that was 1.4-fold higher for the GFP versus transparent hydrogel (1.7 mm⁻¹ and 1.2 mm⁻¹, respectively; data not shown). Together, these results exemplify that the F₀ signal can be strongly affected by the scattering properties of the photosynthetic tissue and the spatial arrangement of light-scattering versus light-absorbing elements in the tissue.

Our results suggest that coral light-scattering modulated (1) the ability of the saturation pulse to fully saturate PSII, (2) the likelihood for actinic effects during ML probing, and (3) the operational volume that is probed during F_v/F_m measurements (Fig. 7). F_v/F_m values for the skeleton hydrogel were greater than 0.1 units (dimensionless) higher than for the transparent and GFP hydrogels when probed with low ML intensities (Fig. 2). This likely indicates that skeleton backscattering facilitated the full saturation of all photosynthetic cells within the tissue volume, creating a homogenous light environment (Fig. 3; Enriquez et al., 2005). In contrast, the steep light gradient in the GFP hydrogels led to rapid attenuation of the saturation pulse light (Fig. 3), leaving only ∼50% of E_d in the lowest layers of the photic zone. In such a scenario, the likelihood of incomplete PSII saturation increases with vertical tissue depth (Serôdio, 2004; Lichtenberg and Kühl, 2015), thus inducing lower F_m values for deeper tissue layers. Optical simulations using Chf-MC showed that increased tissue scattering (from μ_s′ = 1–10 mm⁻¹) reduced the tissue depth for which PSII was fully saturated by >50% (for μ_a = 0.1–1 mm⁻¹; see Supplemental Text; Supplemental Fig. S4B). Chf-MC can serve as an initial point of reference for assessing the likelihood of incomplete PSII saturation in the sample (Supplemental Fig. S4B). Other approaches, including the multiphase flash method, which uses ∼1-s–long multiphase flashes to saturate PSII, could provide additional instrument improvements that reduce the likelihood of incorrect F_m′ estimates (Loriaux et al., 2013).

Optical scattering affected the ML intensity within the photosynthetic tissue and thus the likelihood of ML inducing actinic effects (Supplemental Fig. S5). The relationship between ML used to probe for F and F_m can be nonlinear at higher light intensity settings, leading to a decrease in the F_m/F and thus a reduction in F_v/F_m values (Ting and Owens, 1992). Such nonlinearity is caused by instrument optics, and although this has been tested only for the PAM 101 (Walz; Ting and Owens, 1992), it is likely that the same artifacts contributed to the observed decrease in F_v/F_m for higher ML settings when using the I-PAM system (Fig. 2F).

Light scattering affected the operational volume of the PAM instrument, i.e. the contribution of fluorescence from different vertical tissue depths to observed fluorescence from the sample (Supplemental Figs. S2 and S4A). For samples containing photosynthetic cells with variable intrinsic PS II efficiency, differences in the operational volume could lead to a complex mixture of fluorescence signals from different tissue depths (Fig. 6). Such mixed fluorescence signals could theoretically be decomposed by calculating the depth-specific contribution to observed fluorescence using Chf-MC (Supplemental Text). However, it is a prerequisite that the intrinsic properties of PS II efficiency are known (Klughammer and Schreiber, 2015).

The Ф_PSII and r_ETR differed between the three light-scattering coral mimics (Fig. 4), and the r_ETR of the GFP hydrogel was greater than that of the skeleton and transparent hydrogels when calculated with E_d as a measure of actinic light (Fig. 4C). Using the average E₀ within the photic zone as a measure of actinic light reduced the difference in r_ETR between the GFP and skeleton hydrogels, and the two light curves were identical for E₀ < 400 μmol photons m⁻² s⁻¹. This suggests that for low actinic light levels, measurements of the average in vivo E₀ within the entire photic zone can, to some extent, correct for r_ETR estimates from corals with different degrees of light scattering (Marcelino et al., 2013). For higher actinic irradiance, such correction was not successful, and r_ETR was greater for the GFP hydrogel than for the skeleton hydrogel (Fig. 4D). The enhancement of r_ETR in the GFP scenario was likely due to the presence of the steep light gradient (Fig. 3), ensuring that the irradiance incident on lower cell layers enabled optimal conditions for photosynthesis.

We found that both F_v/F_m and r_ETR were reduced for hydrogels with lower microalgal cell density (Fig. 5, B–D). Lower microalgal cell density decreased light absorption, which enhanced the internal light microenvironment (Fig. 5A) and consequently lowered Ф_PSII. Given that r_ETR is calculated by multiplying Ф_PSII with PAR, differences in the internal light microenvironment can lead to substantial artifacts in the calculation of r_ETR (Fig. 4C). Additionally, correct calculations of absolute ETR require knowledge of the absorption factor, which in itself is affected by optical scattering (Supplemental Fig. S6; Szabó et al., 2014; Wangpraseurt et al., 2014c).

The F_v/F_m measurements were performed for hydrogels with different absorber densities (at ML = 2), which yielded the same F₀ but higher F_m values for the hydrogel with enhanced microalgal cell densities. Although the exact mechanisms underlying these differences are unclear, these first measurements have important implications for coral science, given that microalgal cell density is highly variable between coral species (Drew, 1972) and within a species due to factors such as differences in light acclimation (Falkowski and Dubinsky, 1981), seasonal fluctuations (Chen et al., 2005), and environmental stress (e.g. coral bleaching; Weis, 2008). Bleached corals can exhibit an approximate doubling of the fluence rate within coral tissues compared to healthy corals (Swain et al., 2016; Wangpraseurt et al., 2017a), and such change in the internal light microenvironment was successfully mimicked with our hydrogels with different absorber densities (Fig. 5A). Thus, differences in F_v/F_m and r_ETR between coral individuals with different algal cell densities should be interpreted with caution, and might, to some extent, reflect different optical properties as well as differences in photophysiological status.

PAM-based measurements are often used to assess changes in photochemical efficiency during coral bleaching (Jones et al., 2000; Rodriguez-Roman et al., 2006). During such environmental stress, both the optical properties of the coral and Ф_PSII of the algal symbiont undergo changes over time (Iglesias-Prieto et al., 1992; Wangpraseurt et al., 2017a). Cells from top layers exposed to supra-optimal irradiance are likely to be stressed to a greater extent than cells from deeper layers (Lichtenberg et al., 2016; Wangpraseurt et al., 2016b). We found that changes in cell density and spatial differences in Ф_PSII lead to a misinterpretation of variable fluorescence signals. For instance, hydrogels mimicking stressed corals by containing a top layer with normal algal cell density but with reduced photochemical efficiency, showed much lower r_ETR than hydrogels mimicking bleached corals (see Fig. 1; Table 1 for hydrogel configurations). The high density of the low-performing cells in the top hydrogel layer limited the operational volume in the stressed-coral scenario. In contrast, a reduction in the cell density of top layers enhanced the operational volume to measurements of well-performing lower cell layers, effectively enhancing r_ETR (Fig. 6).

Table 1. Material properties of coral-tissue–mimicking hydrogels.

The algal species was R. salina unless indicated otherwise. The / indicates that no hydrogel layer was fabricated.

Experiment	Hydrogel	Properties	Top Layer	Mid Layer	Base Layer
Coral scattering	GFP	Matrix	ASW+ 1% agarose	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	4%	0%	5%
		Algae	0 cells cm−2	2.5 × 106 cells cm−2	0 cells cm−2
		Thickness	0.75 mm	0.75 mm	2.5 mm
	Skeleton	Matrix	/	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	/	0%	15%
		Algae	/	2.5 × 106 cells cm−2	0 cells cm−2
		Thickness	/	0.75 mm	2.5 mm
	Transparent	Matrix	/	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	/	0%	0%
		Algae	/	2.5 × 106 cells cm−2	0 cells cm−2
		Thickness	/	0.75 mm	2.5 mm
Coral absorption	High	Matrix	/	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	/	1%	5%
		Algae	/	3.5 × 106 cells cm−2	0 cells cm−2
		Thickness	/	0.75 mm	2.5 mm
	Medium	Matrix	/	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	/	1%	5%
		Algae	/	1 × 106 cells cm−2	0 cells cm−2
		Thickness	/	0.75 mm	2.5 mm
Coral stress	Stressed	Matrix	ASW+ 1% agarose	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	0%	0%	5%
		Algae	2 × 106 cells cm−2 (N. oculata)	2 × 106 cells cm−2	0 cells cm−2
		Thickness	0.75 mm	0.75 mm	2.5 mm
	Stressed and bleached	Matrix	ASW+ 1% agarose	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	0%	0%	5%
		Algae	0.3 × 106 cells cm−2 (N. oculata)	2 × 106 cells cm−2	0 cells cm−2
		Thickness	0.75 mm	0.75 mm	2.5 mm
	Healthy	Matrix	ASW+ 1% agarose	ASW+ 1% agarose	ASW+ 1% agarose
		SiO2	0%	0%	5%
		Algae	2 × 106 cells cm−2	2 × 106 cells cm−2	0 cells cm−2
		Thickness	0.75 mm	0.75 mm	2.5 mm

This experimental study has shown that coral optical properties can contrive the interpretation of PAM-based fluorescence measurements. The next step is to develop theoretical models that predict the likelihood of optical artifacts in a sample and ideally correct for such artifacts. We have taken first steps by developing a Chf-MC model for photosynthetic tissues that allows for predicting the likelihood of PSII saturation, the actinic effects of ML, and the operational volume and depth distribution of the collected fluorescence (see Supplemental Text). The simulations can be adjusted to account for absorption of emitted fluorescence. The optical model is limited to a one-layer system, which is best applicable to structurally simple photosynthetic tissues. Future efforts should include the modeling of Chl a fluorescence in multiple tissue layers and in a 3D architecture (Fang, 2010). The use of tissue phantoms with defined optical properties is a promising approach to examine and qualify the precision of variable Chl fluorimetry in plant tissues and biofilms. Quantification of inherent optical parameters in photosynthetic tissues might furthermore lay the experimental basis for better light-propagation models (Jacques, 1998; Mycek and Pogue, 2003; Swartling et al., 2003), which will enable optimal measurement protocols and instrument configurations.

MATERIALS AND METHODS

Experimental Approach

Experiment 1 was aimed at understanding how differences in coral light scattering affect variable Chl fluorescence measurements in identical algal populations, i.e. the same algal culture at identical cell densities. Thick-tissued faviid corals often have light-scattering GFP granules on top of the light-absorbing algal layer, which is also subject to light scattering from the coral skeleton (Lyndby et al., 2016; Wangpraseurt et al., 2017b). The “GFP” hydrogel consisted of a three-layer system with a thin (750 μm) light-scattering upper layer (mimicking GFP scattering), a light-absorbing layer (algae), and a thick light-scattering base layer (skeleton; Fig. 1; Table 1). However, not all corals follow this tissue arrangement, and other corals do not have light-scattering GFP granules. In thin-tissued corals, light scattering can be dominated by the backscattering properties of the skeleton (Enriquez et al., 2005). To mimic this optical configuration, the “skeleton hydrogel” was prepared to be identical to the GFP hydrogel but without the top GFP layer. In the “transparent” hydrogel, the light-scattering skeleton layer was replaced with a 1% agarose gel layer (see schematics in Fig. 1).

In Experiment 2, we examined how changes in coral light absorption might affect variable Chl fluorescence measurements. In healthy corals, algal densities can vary seasonally from ∼1.5–6 × 10⁶ cells cm⁻² (Chen et al., 2005). We therefore created hydrogels with microalgal densities of 3.5 × 10⁶ cells cm⁻² (“high density”) and 10⁶ cells cm⁻² (“medium density”; Fig. 1; Table 1).

Experiment 3 mimicked a coral stress scenario to explore systematically, how combined changes in microalgal density and photophysiology affect variable Chl fluorescence measurements in corals. We mimicked a “stressed coral,” where top algal layers have reduced photosynthetic quantum yields, while lower layers are operating with high yields. For this, we created a light-absorbing algal layer with Nannochloropsis sp. (thickness = 750 μm, algal density = 2 × 10⁶ cells cm⁻²) on top of a Rhodomonas salina layer (thickness = 750 μm, algal density = 2 × 10⁶ cells cm⁻²), exhibiting a F_v/F_m = 0.7. The double layer was placed on top of the light-scattering skeleton hydrogel. We also created a hydrogel mimicking a “stressed and bleached coral” by using a reduced density of Nannochloropsis (3 × 10⁵ cells cm⁻²) in the top layer (Table 1). The “healthy coral” consisted of two layers of R. salina (each 750-μm–thick and with an algal density of 2 × 10⁶ cells cm⁻²) on top of the high-backscatter hydrogel.

Hydrogel Fabrication

To develop a hydrogel with light-scattering properties similar to those of corals, we used a protocol developed for human tissues (Wagnières et al., 1997), where hydrogels with tissue-like properties for visible light were constructed with a reduced scattering coefficient of μ_s′ = 1.5–3.4 cm⁻¹ between 400 and 450 nm (Wagnières et al., 1997). The reduced scattering coefficient of coral skeletons is highly variable, with μ_s′ ranging between 3 and 140 cm⁻¹ (Marcelino et al., 2013; Swain et al., 2016), and the optical properties of living coral tissue apparently exhibit a similar variability (Wangpraseurt et al., 2016a). Given the variability in coral scattering, we did not aim to quantify in detail the reduced scattering coefficient of our hydrogel, but rather to develop a hydrogel that falls within the bulk part of light scattering observed in corals.

The developed hydrogels were composed of (1) a gel-like matrix, (2) light-scattering particles, and (3) light-absorbing algae. We used a 1% agarose (Ultrapure low-melting-point agarose; Thermo Fisher Scientific) solution in filtered (0.2 μm) seawater, which is rather optically clear in the visible part (Wagnières et al., 1997). The agarose was prepared by heating the agarose–seawater mixture in a microwave, ensuring that the solution was clear and free of gas bubbles. The low agarose concentration ensured that the hydrogel was mechanically similar to soft tissues such as coral tissue and exhibited gas diffusion properties similar to seawater. Light scattering was achieved by mixing the hydrogel with defined concentrations of silicon dioxide particles (size fraction: 99% between 0.5 and 10 μm and 80% between 1 and 5 μm; Sigma Aldrich) to achieve the desired scattering (see Table 1; Wagnières et al., 1997). Such silicon dioxide particles are nontoxic to microalgae and cyanobacteria (Dickson and Ely, 2013) and exhibit a good broadband scattering of white light at the chosen particle size distribution (Wagnières et al., 1997).

After adding the silicon dioxide particles, the agarose solution was vortexed for ∼30 s, ensuring a homogenous distribution of the light-scattering particles. The solution was cooled down to ∼30°C, after which the microalgae were added at defined concentrations (see Fig. 1). We selected two types of light-absorbing microalgae, Symbiodinium sp. and Rhodomonas salina. Preliminary experiments were performed with Symbiodinium sp., while the main experiments were performed with R. salina, which was similar in cell size (8–10 μm) and was easier to grow and maintain in a healthy state. For both algal species, PAM-relevant blue light absorption is dominated by Chl a along with additional contributions by Chl c (Kaňa et al., 2013; Wangpraseurt et al., 2014a). However, for the purpose of this study, the type of algal strain is largely irrelevant, as we investigated the effect of basic light-scattering mechanisms on variable Chl a fluorescence measurements. The solution of agarose, SiO₂ and microalgae was transferred rapidly into petri dishes (diameter 35 mm, height 10 mm), where they were left to cure for at least 30 min.

Apparent Optical Properties of Tissue Hydrogels

The light microenvironment of coral mimics was measured in vivo using E₀ microsensors (Rickelt et al., 2016) as described in Kühl (2005) and Wangpraseurt et al. (2012). Briefly, E₀ probes were constructed with a spherical isotropic light-collecting tip of ∼80 μm (Rickelt et al., 2016). The probe was mounted on a motorized micromanipulator (PyroScience) controlled by a PC running dedicated software (Profix; PyroScience) and oriented at an angle of ∼45° relative to the vertically incident light. Measurements of spectral E₀ were performed from the surface of the hydrogel in vertical step sizes of 80 μm. The spectral E₀ was then integrated over the spectral range of PAR (400–700 nm) and expressed in percentage of the incident downwelling irradiance (E_d; Kühl, 2005).

Variable Chl Fluorescence Imaging

We used a variable Chl fluorescence imaging system (Mini I-PAM, Walz; Ralph et al., 2005). The I-PAM was equipped with blue light-emitting diodes (460 nm) and delivered a maximum saturation pulse intensity (SP = 10) of >2,700 μmol photons m⁻² s⁻¹. The ML intensity was calibrated for ML1–12 at frequency 1 using a fast data logger (ULM-500; Walz) connected to cosine-corrected PAR sensor, yielding an average photon irradiance output of 0.3, 0.4, 0.6, 0.8, 1, 1.2, 1.4, 1.6, 1.8, 2.1, 2.3, and 2.6 μmol photons m⁻² s⁻¹. The I-PAM system was mounted on a heavy stand and the hydrogels illuminated vertically from above. Initial measurements were performed to calibrate the focal distance and aperture settings between the camera head and hydrogel samples, after which, focus and aperture were fixed. All measurements were performed with the hydrogels within petri dishes that were placed on top of a black light-absorbing surface. Measurements were performed in a darkened room.

Measurements were performed with dark-acclimated samples (after ∼30 min in darkness) to examine differences in the background fluorescence between the “transparent,” “skeleton,” and “GFP” hydrogels. For each measurement, the PAM settings were fixed (ML = 4, gain = 1, damping = 2, frequency = 1). Measurements were also performed to examine the effects of changes in ML intensity on F₀, F_m, and the calculated F_v/F_m. Each sample was measured at a range of ML intensities (ML settings ranging between 1 and 12), beginning with the lowest ML intensity setting. For each ML intensity, a saturation pulse was applied for 720 ms (intensity setting = 10, yielding 2,700 µmol photons m⁻² s⁻¹), and a resting period of 1 min was used between saturation pulses.

We also examined the effects of light scattering on samples illuminated with defined actinic irradiance levels, i.e. light-acclimated samples. Steady-state light curves of r_ETR versus photon irradiance were measured over a range of actinic incident photon irradiances of PAR (400–700 nm) ranging from 0 to ∼1,600 μmol photons m⁻² s⁻¹. For each light curve, the sample was dark-acclimated for ∼15 min before a light curve was measured using an exposure time of 5 min at each irradiance level. Before starting the steady-state light curves, the ML intensity was adjusted such that F₀ for the different coral mimics yielded comparable values (i.e. F₀ = 0.08). Likewise, ML intensity was adjusted such that F₀ = 0.08 for Experiments 2 and 3 (Fig. 1).

Data Analysis

The effective photosynthetic quantum yield of PSII was calculated as Ф_PSII = (F_m′-F)/F_m′ and relative photosynthetic PSII electron transport rates were calculated as r_ETR = PAR × Ф_PSII (Baker, 2008). Calculated r_ETR versus photon irradiance curves were fitted with an exponential function (Webb et al., 1974) to estimate the maximal relative PSII electron transport rates (rETR_max) and the light-use efficiency factor, α. Nonlinear curve fitting was performed in OriginPro (9.3; OriginLabs) using a Levenberg-Marquart least-squares fitting algorithm.

Optical Simulations

A probability light distribution model was developed to calculate the depth-dependent generation and escape of Chl fluorescence (Chf-MC). Details of the model and optical simulations can be found in Supplemental Text, Supplemental Figures S1 to S6, and Supplemental Table S2.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Schematic of photon energy flow in PAM-based variable Chl fluorescence measurements.

• Supplemental Figure S2. Chf-MC model simulating the penetration of ML.

• Supplemental Figure S3. Chf-MC model investigating the effect of reabsorption on observed fluorescence.

• Supplemental Figure S4. Chf-MC model calculating the penetration depth of the generated fluorescence.

• Supplemental Figure S5. Chf-MC model calculating the percent of photosynthetic tissue overexposed by ML.

• Supplemental Figure S6. Two-layer Monte Carlo simulation of light absorption by microalgal cells for the skeleton hydrogel showing the effect skeletal backscattering on the tissue absorption factor.

• Supplemental Table S1. Abbreviations.

• Supplemental Table S2. The optical properties assumed for the excitation and fluorescence wavelengths..

• Supplemental Text. Chf-MC model to simulate the effect of changes in light scattering and pigment density on calculations of the PSII maximum quantum yield (F_v/F_m).