Going under: The implications of sea‐level rise and reduced light availability on intertidal primary production

Estuarine intertidal flats are known for high rates of primary production, however, globally these ecosystems are under threat from rising sea‐level and accelerated inputs of terrestrial sediments. These stressors simultaneously alter the duration of submersion/emersion periods and water‐column turbidity impacting seafloor photoperiods. To assess how changes in submergence time and water‐clarity can affect benthic primary production, we derived in situ seasonal photosynthesis‐irradiance (P‐I) curves over 2 yr. These were conducted during submerged and emerged periods in a seagrass meadow (Zostera muelleri) and an unvegetated sandflat (dominated by microphytobenthos). P‐I curves showed strong light responses during submergence in both habitats and seagrass during emergence, but in the exposed sandflat there was little relationship between light availability and photosynthesis. Light‐saturated measures of gross primary production (GPPsat) were 2.1 (submerged) and 4.1 times (emerged) higher in the seagrass than the sandflat habitat. Submerged GPPsat was always higher than emerged GPPsat, but this difference was much more pronounced in the sandflat habitat (2.4 vs. 1.3 times for seagrass). Integrating seasonally averaged P‐I curves for both tidal states into a simple model suggests that increasing submergence time could increase primary production in both habitats. However, reductions in water‐clarity will decrease submerged primary production; an effect that will be exacerbated with sea‐level rise. The sandflat trophic state was more sensitive to these environmental stressors because emerged production could not compensate for reductions during submergence. Collectively, regional management of water‐clarity may be key to providing resilience to intertidal habitats, against the inevitable effects of sea‐level rise.


Introduction
Estuarine intertidal soft sediment habitats are among the most productive ecosystems in the world (MacIntyre et al. 1996;Duarte and Chiscano 1999). This productivity is supported by benthic primary producers fuelled by abundant nutrients (sourced from land and stored in sediments) and high seafloor light availability. By oxygenating surface sediment and assimilating nutrients, benthic primary producers alter sediment biogeochemistry, influencing important ecosystem processes such as nutrient regeneration and denitrification, sediment stability, and carbon sequestration (Eyre and Ferguson 2002;Hejnowicz et al. 2015;Hope et al. 2020). Through trophic transfers and export of organic matter, intertidal primary production also helps sustain adjacent coastal ecosystems (Christianen et al. 2017). Changes to the environmental factors that regulate intertidal primary production can therefore have cascading effects on coastal environments. Indeed, a loss of seafloor primary production is already cited as a contributing factor to estuarine regime shifts around the world (e.g., Cooper and Brush 1993;Munkes 2005).
With over 20% of the world's population living within 100 km from the coast (Small and Nicholls 2003), direct and indirect effects of human activities (e.g., dredging and coastal development) are threatening intertidal habitats (Halpern et al. 2007). Globally, anthropogenic stressors are impacting benthic primary production by reducing the extent of intertidal area (e.g., land-reclamation; Airoldi and Beck 2007) and simultaneously degrading seafloor light intensities. Increased inputs of fine suspended sediments from terrestrial runoff (Thrush et al. 2004) and excess nutrients that drive eutrophication (Duarte 1995) are primarily responsible for seafloor light reductions. However, emersion periods on intertidal flats may offer primary producers some resilience to increased water-column turbidity, as it provides a window of high light intensity (Drylie et al. 2018;Mangan et al. 2020b).
Climate change may further magnify the impacts of current intertidal stressors. For example, the inputs of terrestrial sediments and nutrients to the coastal zone will be accelerated due to the predicted increase of severe weather events and associated runoff (Seneviratne et al. 2012). Critically, because intertidal areas are situated at the land-sea boundary, small increases in sea-level rise will alter their extent. This impact may be exacerbated where artificial coastal structures exist, generating coastal squeeze (Pontee 2013). Even in areas of managed retreat, the re-establishment of intertidal habitat in an ecologically relevant time scale is not assured due to the mismatch between predicted sea-level rise and slower evolving morphological responses (Elmilady et al. 2022). Currently, estimates indicate that tidal flats globally are decreasing at a rate of 0.18% per year (Murray et al. 2019) but, with sea-level rise predictions of 1.4 m by 2100 (Rahmstorf 2007), this is expected to accelerate. For example, in New Zealand estuaries, a 1.4 m rise in sea-level could result in an estimated 27-94% loss of estuary intertidal areas (Mangan et al. 2020a). Despite sea-level rise and terrestrial runoff being recognized globally as stressors to coastal environments (Thrush et al. 2004;Halpern et al. 2007), there is little understanding of how varying periods of submergence alongside altered light climates will impact intertidal primary producers.
In temperate intertidal soft sediment environments, two key habitats are seagrass meadows and unvegetated flats that are dominated by benthic microalgae (microphytobenthos). Although microphytobenthos contain less photosynthetic biomass than seagrass per unit area, they can contribute up to 80% of whole-estuary primary production (Christianen et al. 2017;Jones et al. 2017). Microphytobenthos are also a high-quality labile food source (Hope et al. 2020), compared to seagrass (Duarte and Cebri an 1996), and can support a large proportion of coastal benthic consumers (e.g., 74% in the Wadden Sea; Christianen et al. 2017). Despite the recognized importance of these primary producers, only a few in situ studies compare rates of primary production between these key habitats (e.g., Lee et al. 2011;Mishra et al. 2018) and none that address how production may respond to climate related stressors.
Seagrass and microphytobenthos production are strongly influenced by light availability (specifically photosynthetically active radiation [PAR]). In intertidal environments, high rates of primary production have been measured during emerged conditions (Lee et al. 2011;Migné et al. 2018) and where water-clarity is sufficient, during submerged conditions (Drylie et al. 2018;Mangan et al. 2020b). During emergence, light availability is controlled by weather and season, but primary production may also be affected by desiccation and nutrient limitation among other variables (Miles and Sundbäck 2000;Boese et al. 2005;Coelho et al. 2009). In highly turbid estuaries, where benthic primary production is restricted to emerged periods (e.g., Migné et al. 2004;Lin et al. 2020), sea-level rise may reduce this productivity window further. In estuaries with low turbidity, small increases in water depth are likely to have minor effects on seafloor light levels. However, turbidity increases are expected to occur simultaneously with sea-level rise in many places, raising the possibility of synergistic effects on primary production.
Photosynthesis-irradiance (P-I) curves provide a way to assess how primary production varies as a function of light; a proxy for how intertidal systems may be affected by increasing water depth and/or elevated turbidity. Numerous P-I curves have been produced for intertidal primary producers, but many lack a real-world context, using isolated samples (i.e., resuspended microphytobenthos (e.g., Guarini et al. 2002) or individual seagrass blades (e.g., Zimmerman et al. 1991) and often conducted ex situ. Measurements on intact whole-communities are important as they incorporate feedbacks between primary producers and other components of the ecosystem (e.g., macrofaunal bioturbation enhancing benthic production; Lohrer et al. 2016). Furthermore, by considering ecosystem metabolism and production simultaneously, it allows an assessment of trophic state (Eyre and Ferguson 2002). Despite the clear significance of light as a regulator of primary production, we found only five (Clavier et al. 2011(Clavier et al. , 2014Denis et al. 2012;Walpersdorf et al. 2017;Migné et al. 2018) in situ comparisons of intertidal P-I curves across tidal states (submergence/emergence), with only one (Clavier et al. 2014) providing a comparison between seagrass and microphytobenthos dominated habitats. These data are critical to understanding how potential changes in watercolumn turbidity and/or sea-level rise will impact intertidal primary production.
To understand how primary production in two key intertidal habitats is likely to respond to climate related reductions in seafloor light intensity and altered submergence/emergence periods, we measured in situ P-I curves seasonally for 2 yr in a seagrass meadow (Zostera muelleri) and adjacent unvegetated sandflat. Unlike many previous studies, we used large-scale benthic incubation chambers (0.25 m 2 ) to understand how wholecommunity production/metabolism varied with changing light availability and tidal state. Using long-term seasonal measurements, we assessed primary production under a wide range of environmental conditions typical of temperate estuaries to determine, which environmental drivers influenced primary production in these habitats. The P-I curves generated were used to parameterize a simple model to explore how changes in submergence period and light availability may alter gross primary production (GPP) and the trophic state of each habitat.

Flowers et al.
Rising seas, light, and intertidal production

Study site
The study was conducted at Tuapiro Point (37 29 0 5.64 00 S, 175 57 0 13.08 00 E), Tauranga Harbor, New Zealand. Tauranga Harbor is a large (201 km 2 ), barrier enclosed lagoon representative of a globally common estuary type (Stutz and Pilkey 2001). This meso-tidal estuary has a mean water depth of 2.1 m with semi-diurnal tides and a spring-neap tidal range of 1.24-1.62 m (Heath 1975). The intertidal regions (66% of estuary area; de Lange and Healy 1990) consist of two main habitat types; seagrass (Z. muelleri) meadows and bareunvegetated sandflats. The seagrass genus Zostera is common in temperate intertidal environments and are distributed throughout five of the six bioregions of the world (Short et al. 2007). Z. muelleri is New Zealand's sole species of seagrass and is predominantly restricted to intertidal zones (Turner and Schwarz 2006). In Tauranga Harbor, surveys in 2021 estimated seagrass habitat to cover 9% of the intertidal area (Ha et al. 2021).
The study site was located in the mid-intertidal with emersion periods of 4-5 h. Within the site, a plot area of 100 m 2 was marked in the seagrass meadow and adjacent sandflat (> 2 m away from the seagrass fringe). Adjacent plots were used for each seasonal period moving along the seagrass fringe (four plots total); allowing a year gap between sampling within the same plot. The close proximity of the habitats and adjacent plots minimized differences in environmental conditions (i.e., temperature and nutrient availability).

Field sampling
Sampling was undertaken seasonally (every 3 months), for 2 yr between October 2018 (austral spring) and July/August 2020 (austral winter). Due to COVID-19 restrictions, April 2020 was not sampled. To capture the period of highest natural irradiance, each tidal state was sampled around midday high (submerged) or low (emerged) tides across a 2-week period. During two instances, poor weather extended this sampling window to 4-weeks (October 2019 and July 2020). Submerged and emerged sampling was conducted over two consecutive days (one per habitat), with the exception of July 2020 due to poor weather.
Within each habitat, 18-22 benthic incubation chambers (0.25 m 2 ) were deployed in two shore-parallel lines (> 1 m apart; see Supplementary Fig. S1). Chamber bases (L : W : D = 0.5 Â 0.5 Â 0.15 m) were inserted $ 5 cm into the sediment; care was taken to avoid disturbed areas (e.g., ray feeding pits) and to ensure similarities in seagrass cover. The chambers were sealed with Perspex transparent dome lids encapsulating volumes of 34-40 L. Chambers were randomly assigned to one of eight light treatments (0%, 8%, 15%, 30%, 45%, 60%, 85%, and 100% incident light); the 0% and 100% treatments were replicated three times and the remainder in duplicate. Light treatments were achieved by using shade cloth of varying mesh sizes. Within each sampling period, chamber locations and treatments remained the same for both tidal states (unless disturbed).

Gas flux measurements Submerged
Chamber bases, deployed at low tide, were fitted with a PME miniDOT dissolved oxygen logger (1 min sampling interval), a HOBO pendant light and temperature logger (5 min sampling interval) and a Seabird CTD pump (on for 5 s every 45 s; to provide intermittent non-directional stirring of the chamber water) (see Supplementary Fig. S1). While chambers do not incorporate any potential effects of the benthic boundary layer or water movement (via waves/currents) on benthic communities and porewater solute exchange (reviewed by Glud 2008), pumps ensured homogenous mixing of water under consistent flow conditions. An Odyssey PAR sensor (integrated 5 min samples) was also assigned to one replicate of each light treatment (excluding 0% light). Nylon tubing sampling ports (2 m length and 3.2 mm diameter) in each chamber lid enabled seawater extraction. A small inlet on the opposing side ensured the chamber volume remained constant. On the incoming tide, at $ 0.3 m water depth, chamber lids and shade cloth were clamped onto each base, ensuring no air pockets were present. Three pairs of light and dark 1-liter bottles were also filled, sealed and secured at the seafloor to account for water-column production. Once sealed, sampling hoses were flushed before an initial 60 mL seawater sample was extracted from each chamber using a Leur-lock syringe. Three 60 mL ambient seawater samples were also collected. Following an incubation period of $ 4 h, another 60 mL sample was taken from each chamber. Each light and dark water bottle was similarly sampled. Once collected, dissolved oxygen concentrations were measured immediately using an optical sensor (YSI ProSolo ODO/CT) to provide a means of comparing chamber and water-column fluxes.

Emerged
Chamber bases were deployed $ 2.5 h before low tide, once seawater had drained from the site. Prior to incubations, chamber bases were shaded for 30 min to allow the microphytobenthos and seagrass sufficient acclimation to the light treatment (Drylie et al. 2018). Closed-circuit incubations were conducted using a single modified chamber lid fitted with a battery-powered fan, temperature sensor, pressure vent, and an air-in and air-out port connected to a calibrated LI-COR 8100A Automated Soil CO 2 Flux System (following approaches of Streever et al. [1998], Migné et al. [2002], and Drylie et al. [2018]; see Supplementary Fig. S1). The lid was sequentially clamped to each base with incubations lasting 3.5 min; the initial 30 s was treated as a dead-band period for CO 2 flux stabilization (removed prior to analysis). This incubation period was determined following preliminary tests that demonstrated a stable linear change in CO 2 concentration with Flowers et al.
Rising seas, light, and intertidal production minimal variation in humidity/temperature (average variation < 7%). During the incubation, a stream of air was circulated from the chamber to the LI-COR 8100A from which CO 2 (ppm) and moisture content (% humidity) was measured (1 Hz frequency). Once all chambers had been sampled, they were resampled sequentially (2-3 times) until tidal inundation was imminent. In addition to light treatment acclimation, both the duration of submerged incubations and sequential resampling of emerged chambers would have likely captured any potential changes in primary production resulting from vertical migration by microphytobenthos and/or tidal stage (Pinckney and Zingmark 1991).

Site characteristics
Alongside chamber light and temperature measurements, air temperature, humidity and sediment temperature (to 5 cm depth during emerged incubations) was also recorded. Light measurements (Odyssey PAR sensor) were taken at the seafloor during submerged and emerged conditions (referred to as "seafloor PAR"). An additional Odyssey PAR sensor was placed on shore during submerged measures (referred to as "site PAR"), for the determination of water-column light attenuation. To evaluate sediment properties, each day five sediment cores (2.6 cm diameter and 2 cm depth) were collected around each chamber. These were pooled, frozen and stored in the dark for later analysis. After each seasonal sampling event, a 13 cm diameter core was taken to a depth of 15 cm beside each chamber for macrofaunal analysis. This was sieved in situ over a 500 μm mesh and preserved in 70% isopropyl alcohol. For the seagrass habitat, a photograph of each chamber was taken to estimate seagrass percentage cover. To estimate seagrass biomass, a 13 cm diameter core was collected from the center of the chamber, sieved in situ over a 1 mm mesh and frozen within 4 h.

Laboratory analyses
Sediment samples were thawed, homogenized and divided for the analysis of sediment properties. Grain size samples were measured using a Malvern Mastersizer 2000 (particle size range 0.05-2000 μm), following removal of organic matter (10% hydrogen peroxide digestion). Water content was determined from percentage weight loss of samples dried at 60 C, with organic content determined from percentage weight loss on ignition (Heiri et al. 2001). Chlorophyll α (Chl α) and phaeopigment content (μg g dw À1 ) were determined fluorometrically (Turner 10-AU fluorometer), before and after acidification by hydrochloric acid using freeze-dried samples steeped in 90% buffered acetone (Arar and Collins 1997).
As adult bivalves (> 10 mm) have been shown to contribute disproportionally to community metabolism and stimulate primary production (Sandwell et al. 2009;Lohrer et al. 2016), the abundance and size of Austrovenus stutchburyi and Macomona liliana was measured for each sampling event. Seagrass percentage cover was estimated using a 100 random point count analysis (CPCe v 4.1) manually categorized as live blades, dead blades, or unvegetated sediment (Kohler and Gill 2006). Chamber seagrass samples were thawed and separated into above-(leaves) and below-ground (sheath, rhizomes, and roots) biomass and dried at 60 C until a constant weight.

Estimating primary production
Chamber fluxes of carbon (μmol C m À2 h À1 ) were calculated from the submerged dissolved oxygen and emerged CO 2 measurements. For submerged sampling, dissolved oxygen fluxes were calculated using a 10 min average of oxygen concentrations at the start and end of the incubation period. Water column processes accounted for < 5% of the benthic dissolved oxygen fluxes and so were ignored. Dissolved oxygen fluxes were converted into carbon using a photosynthetic quotient of 1.2; a commonly used value for natural communities of both microphytobenthos and seagrass (Ryther 1956). For emerged sampling, fluxes were calculated from the slope of the linear regression of the CO 2 concentration with time (SoilFluxPro v4.0.1). For chambers with multiple measurements during a sampling event, CO 2 fluxes and PAR measurements were averaged.
Dark chambers with 0% light availability provided measurements of sediment community respiration with remaining light treatment chambers (8-100%) providing measures of net primary production. Estimates of GPP were calculated by adding the average sediment community respiration to the net primary production measured in each light chamber.
P-I curves were created using measures of GPP and PAR. For submerged incubations, the average PAR (μmol photons m À2 s À1 ) during the incubation period was calculated from the Odyssey logger within each light treatment chamber. Due to the short duration of emerged incubations, the seafloor Odyssey logger was corrected by the previously determined reductions in light intensity caused by each shade-cloth treatment. P-I curves were created using Eq. 1 (Webb et al. 1974): Where P max is the maximum rate of GPP (plateau of the curve), I is irradiance in PAR (μmol photons m À2 s À1 ) and alpha (α) is the initial slope of the curve; a measure of the photosynthetic efficiency at low light intensity (μmol C m À2 h À1 [photons s À1 ] À1 ). This equation was selected as no evidence of photoinhibition was found. P max and α were estimated using user-specified least squares regression (Statistica v13.0). Consistent with previous literature, R 2 values are presented to describe the strength of the model fit for each P-I curve. The irradiance at which light begins saturating photosynthesis (Ik; μmol photons m À2 s À1 ) was calculated from Eq. 2: Flowers et al.
Rising seas, light, and intertidal production For each sampling date, values of GPP obtained at light levels greater than Ik (i.e., above light saturation) were extracted for comparative analysis (referred to as "GPP sat "). As no GPP values above Ik were recorded for the submerged sandflat in July 2020 and the emerged seagrass in April 2019 due to cloud cover, values from only the 100% light treatment chambers were used.

Statistical analyses
To determine if measures of GPP sat varied temporally (season), between tidal states (emerged and submerged) and/or habitats (seagrass and sandflat), a three-way fixed factor PER-MANOVA was performed (999 permutations). Where significant interactions (p perm < 0.05) occurred, post-hoc pairwise tests were used to identify, which levels differed within each factor.
Distance-based Linear Models were run to identify environmental drivers (e.g., seafloor light intensity, temperature, primary producer biomass, and adult bivalve abundance and size) of GPP sat . Data from all sampling dates, within a habitat/ tidal state were aggregated for this analysis. Where high collinearity (Pearson's r > 0.9) between environmental predictors occurred, the predictor explaining the least variability was removed. All variables were normalized prior to analysis and marginal tests were performed to identify significant individual predictors. Stepwise procedures (using AICc; Burnham and Anderson 2002), then further identified the best combination of predictor variables to create the most parsimonious model. Statistical analyses were completed using PRIMER v7 software with the PERMANOVA+ package on Euclidian distance-based matrices.

Environmental variables
Seasonally, and between habitat types, only minor differences in sediment properties and adult bivalve characteristics were evident (Table 1). Sediment in seagrass and sandflat habitats was fine sand with similar median grain size ($ 175 μm) and mud (< 63 μm; $ 7%), water ($ 26%) and organic matter ($ 2.7%) content. In both habitats, A. stutchburyi occurred in greater abundances than M. liliana ($ 1470 vs. 177 m À2 , respectively). For each species, the mean abundance and size was similar in both habitats.
As a result of weather conditions, temperature, and light intensity (PAR) was variable but seasonal patterns were apparent (Table 2). Higher temperatures and light levels were generally seen during spring/summer, compared to the autumn/ winter. Average temperature and seasonal ranges were higher during emerged conditions (by $ 5 and 10 C, respectively) in both habitats. On average, submergence reduced site light availability at the seafloor by 60% (site PAR vs. seafloor PAR).
No substantial differences in temperature and light regime were detected between habitat types.

Primary producer biomass
Sediment Chl α content (a proxy for microphytobenthos biomass) remained relatively consistent across sampling dates in both habitats and did not differ between tidal states (Table 2). For seagrass, seasonal patterns in percentage cover and biomass were observed in the first year of sampling (spring 2018 to winter 2019), with above-ground and total biomass being highest in summer, before decreasing in autumn and again in winter. During the second year of sampling, seagrass above-ground and total biomass did not vary seasonally and remained relatively low ($ 34 and 110 DW g m À2 , respectively). Seagrass percentage cover was more variable, however consistent with seagrass biomass, a reduction in percentage cover occurred from the first to second year of sampling (by 6-26%). On average, above-ground biomass accounted for 29-41% of total seagrass biomass.

P-I curves
To summarize the relationship between light availability and primary production, data from each seasonal sampling event were pooled and integrated P-I curves for the seagrass and sandflat habitats under submerged and emerged conditions were generated ( Fig. 1; Table 3; see Supplementary  Table S2 for individual P-I curve parameters and Fig. S2 and Table S3 for biomass-corrected seagrass curves). P-I curves showed that GPP during submergence in seagrass and sandflat habitats was strongly driven by light availability (R 2 = 0.75 and 0.69, respectively). However, during emergence, the strength of the modeled relationship remained relatively high in the seagrass (R 2 = 0.45), but this relationship was considerably weaker in the sandflat habitat (R 2 = 0.15). P max was always higher in seagrass compared to the sandflat habitat, and this difference was more pronounced during emerged compared to submerged conditions (3.8 vs. 1.8 times). For both the seagrass and sandflat habitats, higher P max was estimated during submerged compared to emerged conditions (by 1.4 and 2.9 times, respectively). In the seagrass habitat, the saturation irradiance (Ik) was similar during both tidal states (differing by only 8 μmol photons m À2 s À1 ), but in the sandflat habitat Ik was 1.7 times lower during emergence when compared to submergence. Community photosynthetic efficiency (α) was also higher (by $ 2.6 times) in the seagrass compared to the sandflat habitat, and was reduced in both habitats (by $ 1.6 times) during emergence when compared to submergence. This demonstrates that seagrass habitat responded more quickly to increases in PAR and in both habitats, emergence slowed this response.

Seasonal variations in primary production
Consistent with P max , light-saturated GPP sat was always higher in seagrass compared to the sandflat habitat ( Fig. 2 1080 μmol C m À2 h À1 ). Submerged seagrass GPP sat was on average 1.3 times higher than the emerged state, but in the sandflat habitat this difference was much more pronounced (2.4 times). Seasonal trends in GPP sat were also observed. During summer, submerged seagrass GPP sat was greater than in any other season, while emerged seagrass GPP sat was higher in summer and autumn than in either spring or winter (Fig. 2a). Highest rates of submerged and emerged GPP sat also occurred in summer for the sandflat, with lowest rates occurring in winter during submergence (Fig. 2b). Light-saturated net primary production was always positive in each habitat type and tidal state indicating autotrophic conditions (i.e., more carbon fixation than community respiration; see Supplementary Fig. S3  and Table S4).

Environmental predictors of production
Several environmental predictor variables were correlated with measures of GPP sat in marginal tests (see Supplementary  Table S5), but few contributed significantly to the stepwise models (Table 4). Environmental predictors explained between 44 (submerged) and 63% (emerged) of the variation in GPP sat for seagrass habitat. In submerged conditions, PAR was the single greatest contributor (35%) to the variance in seagrass GPP sat, whereas during emergence, temperature was a more important predictor (48%). Above-ground seagrass biomass contributed a further 15% of the total explained variance in emerged conditions. In the sandflat habitat, environmental predictors explained between 19 (emerged) and 48% (submerged) of the variation in GPP sat , with temperature the most significant predictor in both tidal states. (See Supplementary  Tables S5 and S6 for marginal tests and stepwise model results for light-saturated net primary production and sediment community respiration).

Discussion
Seafloor light availability is currently threatened by anthropogenic activities that are driving increasing inputs of terrestrial sediment into coastal waterways (Thrush et al. 2004). To investigate the consequences of reduced seafloor light availability on intertidal primary production, this study obtained in situ seasonal measurements of P-I curves over 2 yr during submerged and emerged conditions in a seagrass meadow and an adjacent unvegetated sandflat. Owing to a substantially greater photosynthesizing biomass, the seagrass meadow consistently maintained higher rates of light-saturated GPP sat compared to the sandflat (as demonstrated by Drylie et al. 2018;Lin et al. 2020). Additionally, primary production was reduced in both habitats during emergence, as environmental factors such as desiccation (Boese et al. 2005;Coelho et al. 2009), nutrient availability (Miles and Sundbäck 2000), and seagrass self-shading (Clavier et al. 2011) likely limited production.
By undertaking comparable measurements during submergence and emergence, using the seasonally averaged P-I curves we can estimate daily rates of GPP for our intertidal habitats that account for tidal state. Assuming a constant annual average incident PAR of 750 μmol photons m À2 s À1 (measured previously at the site; Mangan et al. 2020a), a 60% reduction in incident light availability during submergence (Table 2), and a 50% submergence period for a 12 h daily photoperiod, daily GPP was estimated to be 56 and 21 mmol C m À2 d À1 for the seagrass and sandflat habitats, respectively. Of the daily GPP, the contribution of submerged and emerged production in the seagrass habitat was approximately equal (53% vs. 47%, respectively), while in the sandflat habitat, the contribution 11.0 AE 1.2 11.6 AE 0.7 Note: Data are arranged by austral season for ease of interpretation. PAR, photosynthetically active radiation; SUB, submerged; EMG, emerged; AG, above-ground; DW, dry weight; Chl α, sediment chlorophyll α content.

Flowers et al.
Rising seas, light, and intertidal production by submerged production was double that compared to the emerged production (67% vs. 33%). Our daily GPP estimates were identical to the seasonally averaged integrated GPP sat measurements (Fig. 2) and consistent with the literature. For example, our sandflat daily GPP estimate was similar to that observed by Migné et al. (2004) (25 mmol C m À2 d À1 ) and the mean value reported by Cahoon (1999) based on a review of intertidal microphytobenthos production (26 mmol C m À2 d À1 ). Using P-I curves corrected for seagrass biomass (see Supplementary  Fig. S2 and Table S3), our daily GPP estimate for the seagrass habitat (1.3 mmol C g À1 DW seagrass m À2 d À1 ) was also within the range reported by Duarte et al. (2010) that reviewed 13 seagrass species (0.2-10.7 mmol C g À1 DW seagrass m À2 d À1 ). Similarly, there is some empirical evidence that our P-I curve parametrizations are broadly applicable. Primarily, the saturation irradiance (Ik) values of the P-I curves are consistent with those previously reported in published literature. For the submerged sandflat, Ik was identical to the median value (258 μmol photons m À2 s À1 ) reported by Mangan et al. (2020a) that reviewed 42 P-I curves from wholecommunity intertidal unvegetated flats. Similarly, the seagrass Ik during both tidal states (192)(193)(194)(195)(196)(197)(198)(199)(200) μmol photons m À2 s À1 ) is comparable to the mean Ik (194 μmol photons m À2 s À1 ) of nine seagrass populations (seven species) reported by Vermaat et al. (1997).
In a global context, our study site would be considered oligotrophic and in a healthy state with abundant shellfish, extensive seagrass meadows and clear water that supports high rates of submerged primary production (Vieillard et al. 2020). Our seasonally averaged P-I curves demonstrate that GPP in seagrass meadows and unvegetated sandflats are strongly related to seafloor light availability during submergence. However, this relationship weakened (i.e., lower R 2 values) during emergence, especially in the sandflat. Oligotrophic systems are underrepresented in intertidal whole-community P-I curve studies, which have focused on muddy (and turbid) unvegetated flats where production is limited to emerged  Table 3 for parameter estimates). Curves were constructed from seven seasonal sampling events (austral spring [October] 2018-winter [July/August] 2020). Note the difference in Y-axis scale between seagrass and sandflat habitats and X-axis scale between submerged and emerged conditions. conditions (Migné et al. 2004;Lin et al. 2020). Unlike mudflats, oligotrophic sandflats do not retain high sediment water content and therefore rarely develop microphytobenthos biofilms. This, in turn, can reduce emerged primary production in sand dominated systems as microphytobenthos are more likely to be limited by factors such as desiccation and/or nutrient availability (Miles and Sundbäck 2000;Coelho et al. 2009;Lin et al. 2021). Integration of data from sandy, oligotrophic systems, and under different tidal states, will therefore have important implications for global intertidal carbon budgets (e.g., those estimated by Lin et al. 2020).
While P-I curves illustrate the potential effects of changing light climate on benthic primary production, our short-term flux measurements do not capture the potential for adaptation to changing light conditions (i.e., changes in pigment/ microphytobenthos community composition; Du et al. 2009;Kohlmeier et al. 2014). Across sites of increasing water-column turbidity, Drylie et al. (2018) found that seagrass (Z. muelleri) meadows were able to up-regulate production during emergence to compensate for decreased submerged production, indicating we may have underestimated the degree to which seagrass ecosystems can compensate for long-term reductions in submerged light availability. However, Drylie et al. (2018) observed no such upregulation of emerged GPP with increasing turbidity in microphytobenthos dominated sandflats. The sensitivity of sandflat microphytobenthos to reduced light climate during submergence, and a lack of capacity to compensate during emergence, may have considerable implications for coastal foodwebs if sea-level rise is accompanied with increased turbidity.
Our seasonally averaged submerged sandflat P-I curve indicates that, at $ 300-400 μmol photons m À2 s À1 , further decreases in PAR strongly affect GPP (Fig. 1), indicating a threshold correlated with observed shifts in the structure and functioning of intertidal sandflats (Thrush et al. 2021;Gammal et al. 2022). A recent New Zealand wide and globally unique experiment conducted on unvegetated sandflats (24 sites across 15 estuaries, spanning 12 of latitude), demonstrated that benthic ecological interaction networks in turbid systems (daily averaged submerged PAR > 350 μmol photons m À2 s À1 ) were relatively simple, had fewer connections and no feedbacks when compared to sites with clearer water (PAR > 420 μmol photons m À2 s À1 ; Thrush et al. 2021). Furthermore, from the same experiment, Gammal et al. (2022) demonstrated shifts in biodiversity-ecosystem functioning relationships when daily submerged PAR was < 420 μmol photons m À2 s À1 . At lower light levels, macrofaunal biodiversity loss reduced ecosystem multifunctionality; a relationship not seen in clear estuaries. Although, it is recognized that elevated turbidity may impact other ecosystem components (e.g., suspension-feeders; Hewitt and Norkko 2007), the correlation between light limitation of microphytobenthos and broader changes in the ecosystem emphasizes the need to improve coastal light climates.
Alongside changes in light availability, additional environmental variables were identified as important drivers of benthic primary production above light saturation. Notably, temperature was identified as an important predictor of GPP sat in both habitats, illustrating the positive effect it has on metabolic rates. Under future climate scenarios, air and water temperatures are expected to increase (Ruela et al. 2020;IPCC 2021). However, there are limits to the optimal temperature range for production; 16-35 C in seagrass (reviewed by Lee et al. 2007) and 20-36 C in sandflats (Rasmussen et al. 1983;Migné et al. 2004). Proportionally, temperature can have a greater impact on sediment community respiration than primary production (Hubas et al. 2006;Lee et al. 2011). Consequently, this can result in an ecosystem becoming more heterotrophic (i.e., community respiration exceeds GPP), but mixed results have been observed (Alsterberg et al. 2011;Burkholz et al. 2019).
Despite both sea-level rise and elevated inputs of terrestrial sediment being described as major threats to coastal ecosystems worldwide (Thrush et al. 2004;Halpern et al. 2007), little is known about the combined impacts on the productivity of intertidal flats. In order to evaluate how the current daily integrated GPP may be affected by simultaneous changes in Table 3. Seasonally averaged P-I curve parameters (AE 95% CI) for seagrass and sandflat habitats under submerged and emerged conditions. Parameter ranges from individual sampling events are presented in parentheses. Flowers et al. Rising seas, light, and intertidal production submergence periods (caused by rising seas) and seafloor light availability (a proxy for turbidity and/or sea-level rise), we devised a simple conceptual model integrating the seasonally averaged P-I data ( Fig. 1; Table 3). The model suggests that, provided the water-column remains clear, current integrated GPP should increase with moderate sea-level rise in both habitats (Fig. 3a,b). This is because exposure to environmental factors that limit production during emergence is reduced. As seafloor light climate during submergence declines, the emerged period contributes a greater proportion of the integrated GPP. However, sea-level rise reduces this productivity window, further accelerating the decline in GPP. There are also important differences between habitats both in terms of absolute rates of integrated GPP and their sensitivity to sealevel rise and reducing light climate. Importantly, seagrass meadows are less sensitive to increasing water-column turbidity because they can maintain higher rates of production during emergence than sandflat microphytobenthos (e.g., Drylie et al. 2018;Lin et al. 2020). Integrated GPP in unvegetated habitats is therefore expected to decline more rapidly in response to reduced light levels during submergence, compared to seagrass meadows (Fig. 3a,b). and emerged (white bars) conditions. Bars (mean + 1 SE; n = 3-15) arranged by austral season. See Supplementary Fig. S3 for light-saturated net primary production (NPP sat ) and sediment community respiration data. Photosynthesis to respiration ratios (Eyre and Ferguson 2002) provide an estimate of an ecosystems trophic state; values > 1 indicate systems that are net autotropic (i.e., GPP exceeds community respiration) and those < 1 heterotrophic. Using the seasonally averaged P-I curves and sediment community respiration rates (see Supplementary Fig. S3) with the assumption of a 12 h light and dark period, we estimated how trophic status varied with light and submergence period (Fig. 3c,d). The seagrass habitat maintains an autotrophic state over a greater range of light and tidal state conditions than the unvegetated sandflat; a function of the higher rates of GPP and ability to maintain production during emergence. Conversely, the sandflat habitat was net heterotrophic under most conditions (especially as the light climate worsened), and became less heterotrophic (i.e., higher p/r ratios) as emergence period increased. This somewhat counterintuitive result (sandflat GPP is much lower during emergence), arises because emerged sediment community respiration is substantially lower, likely due to reduced macrofaunal activity (see Supplementary Fig. S3 and Table S4). Although seagrass habitats may remain net autotrophic over a greater range of conditions, this does not necessarily translate to a greater resilience for intertidal foodwebs. In temperate coastal foodwebs, few macrofauna feed directly on seagrass blades due to its low food quality and digestibility; seagrass primarily enters the detrital foodweb (Duarte and Cebri an 1996;Vizzini et al. 2002), whereas microphytobenthos are highly labile and often directly consumed Fig. 3. Variation in daily integrated GPP (a and b) and production to respiration ratios (p/r; c and d) for intertidal seagrass (a and c) and sandflat (b and d) habitats. Response surfaces are plotted as a function of submergence time (to represent sea-level rise [1 = permanently submerged]) and the proportion of incident PAR (750 μmol photons m À2 s À1 ) reaching the seafloor during submergence (a proxy for turbidity and/or sea-level rise). p/r ratios > 1 indicate the habitat is net autotrophic whereas values < 1 are heterotrophic. Plots were created using seasonally averaged P-I curves and sediment community respiration values derived under submerged and emerged conditions (see text for details). Black squares indicate an estimate of the current seasonal average daily GPP and p/r. Note the difference in GPP Y-axis scales between habitats.

Flowers et al.
Rising seas, light, and intertidal production (Hope et al. 2020). This indicates that temperate coastal foodwebs may be more sensitive to changes in microphytobenthos production/biomass than seagrass. Our conceptual model (Fig. 3) contextualizes some of the potential changes in benthic primary production as seafloor light intensity and submergence times change. However, climate change and anthropogenic activities could drive interactions with other simultaneously changing environmental variables and further impact benthic primary production. For example, differences in macrofaunal community structure have strong feedbacks to primary production and ecosystem function (Sandwell et al. 2009;Pratt et al. 2014;Lohrer et al. 2016). Similarly, long-term inputs of terrestrial sediment not only elevates water-column turbidity but also increases sediment mud content (Thrush et al. 2004). This can alter macrofaunal communities (e.g., reduction in diversity [Pratt et al. 2014]) and negatively influence primary production (Thrush et al. 2003;Pratt et al. 2014). These examples serve to highlight the complexities of system-level interactions in a rapidly changing world.
Globally, intertidal regions have been or will be affected by changes in water-column turbidity and sea-level rise to some degree (Thrush et al. 2004;Murray et al. 2019). Future research identifying how soft sediment ecosystems respond to changes in light availability will therefore be invaluable for the management of these threats. Our results emphasized that emerged production may not offer much resilience to increasing water-column turbidity, particularly for microphytobenthos dominated sandflats. For high turbidity intertidal habitats already reliant on emerged primary production (e.g., Migné et al. 2004;Lin et al. 2020), sea-level rise represents a significant threat. However, we demonstrate that in more oligotrophic systems, benthic productivity can remain high when the water-column remains clear (Clavier et al. 2011;Mangan et al. 2020b). This highlights that, although a global problem, localized management of coastal water-quality will play a crucial role in providing resilience to benthic primary producers as sea levels rise.

Data availability statement
Data is available in article Supplementary Information and raw data is available on request from the authors.