Thalassia testudinum Banks ex König (turtle grass) is the dominant seagrass species in the subtropical/tropical Atlantic and Caribbean region (den Hartog, 1970; Green and Short, 2003). This species has high light requirements and is considered to be the climax species in marine seagrass systems (Zieman, 1982). T. testudinum is also the dominant seagrass and dominant physical structure in Florida Bay, a shallow (<2 m, Schomer and Drew, 1982) subtropical estuary that exhibits wide variations in salinity and water clarity (Zieman, 1982; Hall et al., 1999; Durako et al., 2002). Two massive die-offs of T. testudinum have occurred in Florida Bay, United States, the first in 1987–1990 (Robblee et al., 1991) and more recently in late summer 2015 (Hall et al., 2016). Several years after the 1987–1990 die-off, Florida Bay began exhibiting widespread turbid conditions due to phytoplankton blooms and resuspended sediments (Phlips and Badylak, 1996; Stumpf et al., 1999). This increase in turbidity was associated with additional light-stress-induced losses of T. testudinum in western and central Florida Bay (Durako et al., 2002). The two die-off events were nearly equivalent in terms of size, locations, seasonality, speed and appearance (Hall et al., 2016) suggesting common underlying causative factors. Hypersalinity, elevated water temperatures and water column stratification leading to bottom-water anoxia, hypoxic stress and sulfide toxicity are thought to have lead to increased stress and subsequent mortality of very dense T. testudinum (Hall et al., 2016).

In response to the 1987–1990 die-off, the Comprehensive Everglades Restoration Program (CERP) was initiated to increase freshwater inflow to Florida Bay back to historical patterns (Herbert et al., 2011). Decreased salinity is a major management goal of CERP, which may affect seagrass species recovery patterns. Fourqurean et al. (2003) predicted that the more estuarine seagrasses, such as Halodule wrightii and Ruppia maritima, would increase in abundance at the expense of the marine T. testudinum. Because seagrasses are the dominant benthic community in Florida Bay, changes in their species distributions and recruitment patterns are being used as a central performance measure to assess CERP success.

Seedlings are an important source of recruitment in recovery of T. testudinum, following disturbance events (Whitfield et al., 2004). Because T. testudinum exhibits continuous embryo development, germinating as embryos are released from parent plants, there is no seed bank (Kuo and den Hartog, 2006). Thus, seedlings physiological health, survival and growth are dependent on the environmental conditions existing during the release of the annual cohorts. Because T. testudinum exhibits monopodial clonal growth (Tomlinson and Vargo, 1966), the use of field-collected short shoots rather than seedlings in experimental studies could result in pseudoreplication. This makes seedlings ideal independent experimental units on which to assess the effects of light and salinity. In a previous examination of diurnal variation in chlorophyll fluorescence of T. testudinum seedlings in response to variations in salinity and light, we observed significant reductions in photochemical efficiency of photosystem II (F_v/F_m) under hypersaline (salinity 50), but not hyposaline (salinity 20) conditions, suggesting a greater sensitivity to high-salinity stress (Howarth and Durako, 2013a). In contrast, significant increases in F_v/F_m in shaded seedlings indicated low-light acclimation. Seedlings in the salinity 50 treatment also had significantly reduced chlorophyll and carotenoid contents and increased non-photochemical quenching (NPQ) compared to the salinity 20 and 35 treatment seedlings, while shading resulted in increases in chlorophyll b resulting in significant reductions in chlorophyll a:b ratios (Howarth and Durako, 2013b).

The measurement of spectral reflectance [R(λ)] is a non-destructive optical technique that has long been used as an indicator of plant stress (Carter and Knapp, 2001). Leaf R(λ) is an apparent bio-optical property that varies largely in response to changes in leaf pigments (Sims and Gamon, 2002). There are several spectral-reflectance indices based on reflectance at specific regions of the spectrum, which may be used to predict relative leaf pigment contents (Sims and Gamon, 2002). Changes in pigments may reflect plant stress and affect the ability of seagrasses to tolerate variations in light through the regulation of energy transduction (Dennison and Alberte, 1986; Beer et al., 2000; Ralph et al., 2002; Belshe et al., 2007). For example, carotenoids may contribute to photosynthesis as accessory pigments or have photoprotective functions (Ralph et al., 2002). Under conditions of excess irradiance and hypersalinity, carotenoids that comprise the xanthophyll cycle will dissipate excess energy, which avoids damaging photosystems (Demmig-Adams et al., 1996; Howarth and Durako, 2013b). Anthocyanins may also protect leaves from excess visible or UV light or they may serve as antioxidants (Gould et al., 2002). Because pigments serve either photosynthetic or photoprotective functions, changes in R(λ) may correspond with changes in relative photosynthetic efficiency and capacity under various environmental conditions (Bargain et al., 2013). Pigment reductions and increased spectral reflectance are typical responses to plant stress (Carter and Knapp, 2001). However, turbid low-light conditions can result in increases in leaf chlorophyll content and a reduction in the chlorophyll a to b ratio (Dennison and Alberte, 1985; Longstaff and Dennison, 1999; Cummings and Zimmerman, 2003). These changes in chlorophyll contents increase the amount of light absorbed for photosynthesis and thus, reduce reflectance, especially in the blue (400–500 nm) and red (660–680 nm) regions of the spectrum (Sims and Gamon, 2002). In addition to being non-destructive, measurement of R(λ) is rapid and can be applied across multiple spatial scales (Gamon and Qui, 1999), which is why it has been widely used as an indicator of plant health for terrestrial plants (Carter and Knapp, 2001; Sims and Gamon, 2002). In contrast, this methodology has had very limited application in quantifying stress responses in seagrasses (Thorhaug et al., 2006). Here, we report on variations in leaf spectral reflectance of the T. testudinum seedlings in response to 2 weeks exposure to three salinity treatments (20, 35, and 50) and two light treatments (full sun and 50–70% reduction from ambient light). Based on our previous observations on photochemical efficiency and pigment contents (Howarth and Durako, 2013a,b), we hypothesized that R(λ) would increase in response to hypersalinity, but decrease under shaded conditions.

Normalized reflectance spectra for the 20 and 35 salinity treatment leaves were typical of green vascular plant leaves containing chlorophylls a and b (Figure 2). Reflectance was relatively low and constant from 400 to 500 nm then increased toward a peak around 550 nm followed by a decrease to another low around 675 nm and then rapidly increasing to 700 nm. There was an increase in variability among spectra as salinity increased (Figure 2) and a flattening of spectral reflectance at the highest salinity (Figure 3). Shade-treatment seedlings generally exhibited higher reflectance between 525 to 575 nm and lower reflectance around 675 nm compared to the unshaded (Sun) treatment.

Multivariate (nMDS) ordination indicated that hypersalinity had a greater effect on spectral reflectance than hyposalinity or shading (Figure 4). There was clear separation of the hypersalinity treatment seedlings along the primary ordination axis while most of the lower two salinity treatment samples were clustered. All of the Sun 50 replicates and 3 of the 4 Shade 50 replicates are separated from the other treatments on the right side of the ordination plot. Two-way crossed ANOSIM indicated that reflectance spectra for Sun and Shaded treatments were not significantly different (global R = -0.028, p = 0.569). In contrast, salinity had a significant effect on spectral reflectance (global R = 0.458, p = 0.001), with both the salinity 20 (R = 0.682, p = 0.007) and 35 (R = 0.464, p = 0.014) treatments being significantly different from the salinity 50 treatment, but not different from each other (R = 0.224, p = 0.07).

Similarity percentage analysis of the salinity data with light treatments pooled indicated that no single wavelength contributed more than 1% to the dissimilarity among salinity treatment reflectance spectra (Figure 5). Consistent differences among all three salinity treatments occurred across the green wavelengths (530–580 nm), with additional discrimination between 20 versus 50 and 35 versus 50 treatments in red wavelengths (630–690 nm). The Red:Green reflectance ratio was significantly affected by salinity (H = 11.58, df = 2, p = 0.003), being highest in the salinity 50 treatment and not significantly different between the salinity 20 and 35 treatments (Figure 6A). The photochemical reflectance index (PRI) was also significantly affected by salinity (F_2,23 = 8.22, p = 0.02) being lowest in the salinity 50 treatment and similar between the salinity 20 and 35 treatments (Figure 6B).

Leaf spectral reflectance [R(λ)] for T. testudinum seedlings exhibited a greater response to increased salinity than to low salinity or light reduction, after 2 weeks exposure. Despite a 50–70% reduction in ambient irradiance, daily mean PAR quantum flux for the Shade treatments was approximately 8 mol quanta m^-2 day^-1, which exceeds the daily light requirements estimated for tropical seagrasses (Chartrand et al., 2012, 2016). R(λ) in the Shade treatments exhibited only slight increases in the green wavelengths and small decreases in the red wavelengths. These changes are consistent with the small, but non-significant increases in chlorophyll a and b contents in the shaded seedlings relative to the full sun treatments (Howarth and Durako, 2013b). In contrast, seedlings exposed to salinity 50 conditions for 2 weeks exhibited a distinctive flattening of their R(λ) spectra, with reduced reflectance in the green and increased reflectance in the red. This resulted in a significant increase in the Red:Green ratios in the hypersalinity treatment relative to the control and hyposalinity treatments. Increased Red:Green ratios generally correlate with elevated anthocyanin contents (Gamon and Surfus, 1999). Leaf reddening due to increased anthocyanins is associated with numerous stressors, such as temperature extremes and high UV/visible radiation, in terrestrial plants (Chalker-Scott, 1999). Anthocyanin-induced leaf reddening has also been observed in 15 seagrass species (Novak and Short, 2010). Leaf reddening in T. testudinum is associated with high-light environments (Novak and Short, 2011). Although no differences in growth were observed, red-leafed T. testudinum shoots have shorter and narrower leaves than green-leafed shoots and higher effective quantum yields (Novak and Short, 2011). Elevated anthocyanins may function as photoprotectants or antioxidants, allowing seagrasses to maintain higher effective quantum yields after exposure to one of more stressors (Novak and Short, 2010). The significant decreases in PRI values in the salinity 50 treatment indicate reduced epoxidation of xanthophyll cycle pigments and reduced photochemical efficiency (Gamon et al., 1992). These changes in the R(λ) spectra and the two reflectance indices for the salinity 50 seedlings were consistent with the reduced leaf chlorophyll and carotenoid contents in these leaves (Howarth and Durako, 2013b) and the observed significant reductions in F_v/F_m (Howarth and Durako, 2013a).

After 14 days at target salinities, R(λ) spectra, Red:Green ratios and PRI values for the salinity 20 and 35 treatment seedlings were not significantly different, reflecting similar pigments contents and photochemical efficiencies of the leaves in these two treatments (Howarth and Durako, 2013a,b). Our results are unlike a previous report of significant changes in R(λ) spectra for T. testudinum after 24 h exposure to hyposaline (16 versus 32 salinities) conditions (Thorhaug et al., 2006). These differing results may indicate that a tolerance threshold may exist for T. testudinum between salinities of 20 and 16. This should be tested as it could have significant management implications with respect to freshwater releases associated with CERP. Alternatively, the differing results may reflect different methods of exposure to the hyposaline conditions. Thorhaug et al. (2006) placed plants, which had been acclimated to a salinity of 32 for several days, directly into the salinity 16 treatment. This simulated a “pulsed” exposure to hyposalinity. In contrast, we gradually exposed seedlings, which were acclimated in flow-thru vaults for 4 weeks in salinities of 29–35, to changing salinities of 2 per day until our target salinities were reached. Previous studies on the seagrass Halophila johnsonii, which placed plants directly in treatment salinities within 24 h of collection (”pulsed” exposure) reported high mortality in both hyper- and hypo-salinity treatments (Torquemada et al., 2005; Kahn and Durako, 2008; Gavin and Durako, 2012). In contrast, gradually reducing or increasing salinity can extend seagrass salinity tolerances by approximately a salinity of 10 (Kahn and Durako, 2006; Koch et al., 2007; Griffin and Durako, 2012). Thus, the rate of change seems to be an important consideration in the resilience of seagrasses to environmental stresses.

Our observations indicate that leaf spectral reflectance is a sensitive indicator of plant stress in T. testudinum seedlings. The results of multivariate analyses of reflectance spectra from 400 to 700 nm and statistical comparisons of two reflectance indices all suggested greater sensitivity of seedlings to hypersalinity than to hyposalinity. These results are highly consistent with previous measurements of chlorophyll fluorescence (Howarth and Durako, 2013a) and pigment content in these seedlings (Howarth and Durako, 2013b). The occurrence of extreme hypersaline conditions during the warmer seasons is thought to have played a major role in the two massive die-offs of T. testudinum in Florida Bay (Hall et al., 2016). Our data suggest that seedlings of this species are also highly susceptible to hypersalinity stress. Because spectral reflectance measurements are rapid, non-invasive, and have great potential use in remote sensing, they should be more widely applied for assessing stress responses of seagrasses to changing environmental conditions.

MD and JH together designed and executed the research project. MD and JH conducted the spectral reflectance measurements. MD lead the reflectance data analysis and drafted the manuscript with the assistance of JH. All co-authors commented on and approved the final manuscript draft.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.