Introduction

Quantification of an ecosystem’s carbon balance is pivotal for understanding both ecosystem functioning and the global carbon cycle. The lack of data regarding carbon budgets of coastal environments and their components (e.g. Gattuso et al. 1998) limits our understanding of the role of these environments in the global carbon cycle (Borges et al. 2005). This is particularly true for macroalgal beds, which are more effective carbon sinks than planktonic ecosystems (Pessarrodona et al. 2022) because of their higher biomasses and longer turnover times (~1 year for macroalgal communities vs. <1 month for oceanic plankton) (Smith 1981). Macroalgal communities thus have a significant role in the global carbon cycle (Duarte 2017; Pessarrodona et al. 2022), comparable to that of forests on land.

As such, previous studies have evaluated macroalgal metabolism mostly as biomass accumulation (Pessarrodona et al. 2022) and less frequently from a physiological perspective, performing measurements over thalli pieces up to entire organisms and scaling fluxes over biomass or tissue surface area (e.g. Randall et al. 2019). Yet, the emergent properties displayed by communities result in different responses to excess light (Binzer and Middelboe 2005; Tait and Schiel 2011; Migné et al. 2021), temperature (Tait and Schiel 2013), desiccation (Hunt and Denny 2008) or increased CO2 (Sand-Jensen et al. 2007) than for individual organisms. An ecological perspective on the assessment of the carbon budget of macroalgae should thus focus on communities, scaling fluxes over unit area of seafloor (Bordeyne et al. 2020).

In the intertidal zone, macroalgae are subjected to large variations in abiotic factors (Newell 1979). The vertical zonation of intertidal macroalgae is in part constrained by the effects of abiotic factors on their metabolism (Newell and Pye 1968). As such, metabolism of intertidal macroalgae has often been studied either under immersion only (Middelboe et al. 2006) or in response to desiccation specifically (Bell 1993). Yet, the dichotomy between the “benign” low-intertidal zone and the “stressful” high-intertidal zone is an oversimplification (Tomanek and Helmuth 2002). Species that spend substantial portions of their lifetime emerged are necessarily adapted to these conditions (Nicastro et al. 2010; Tagliarolo et al. 2012). This is true of intertidal macroalgae, which can display high metabolic rates during exposure to air (Maberly and Madsen 1990; Madsen and Maberly 1990) especially when adapted to longer exposure periods (Quadir et al. 1979; Migné et al. 2015b). Thus, macroalgal production of intertidal species is physiologically meaningful during emersion. Resistance mechanisms protect macroalgae during emersion periods, even if at the cost of a reduced activity. Resistance of macroalgal thalli for example to excess of light during emersion is provided by their xanthophyll cycle, which limits or prevents the destruction of photosystem II at a metabolic cost but with the ability to recover pre-stress yields as soon as conditions are favourable again (Gévaert et al. 2003). Similar responses have been shown for desiccation due to changes in either temperature (Bell 1993; Hunt and Denny 2008) or relative humidity on tissue surface (Kawamitsu and Boyer 1999). In canopy-forming macroalgae, the stress of aerial exposure may also be limited through self-shading, preventing algae from overheating, desiccation and photoinhibition (Migné et al. 2021). As a consequence, estimating metabolism of intertidal species during emersion turns out to be significant for an accurate evaluation of their contribution to the carbon budget.

For an accurate evaluation of the contribution of macroalgal beds to the carbon budget, integrating the variations of abiotic factors at various temporal scales (Schagerl and Möstl 2011; Bordeyne et al. 2017) is needed. This is especially so for intertidal communities where light and tide interact at different temporal scales through oscillating patterns. These oscillations likely subsidise carbon fluxes (Steever et al. 1976; Odum et al. 1995) in several ways. During immersion, natural variations in irradiance enhance photosynthesis of understory macroalgae (Tait and Schiel 2011). Across the tidal cycle, high irradiance during emersion enhances gross primary production with respect to immersion (Bordeyne et al. 2017). As a consequence, carbon fluxes of intertidal communities during emersion should be specifically studied considering both the integrated effects of abiotic factors during the duration of the emersion period as well as the effect of light at finer temporal scales.

Studies on emersed rocky shore ecosystems dominated by fucoid canopies reveal high production from hourly metabolism measurements (Golléty et al. 2008; Bordeyne et al. 2015), suggesting that annual carbon budgets of these ecosystems should be autotrophic. We thus hypothesised that the annual carbon budget of an intertidal shore dominated by the canopy-forming macroalga Ascophyllum nodosum is autotrophic during emersion. The aims of the present study thus are first, to evaluate this hypothesis and second, to identify the regulating factors of this budget at the scale of the community. For this purpose, we built upon previous findings (Golléty et al. 2008) to better integrate the effects of abiotic factors at different temporal scales. In particular, CO2 fluxes were measured over whole periods of emersion at different times of the year to establish relationships between primary production or respiration and emersion duration. CO2 fluxes were also measured under artificially varying irradiance to establish relationships between primary production and irradiance. The photosynthetic parameters thus derived from these production–irradiance (P–I) curves were used, as shown previously (Falkowski and Raven 2007), to predict rates of primary production at different irradiances. Hourly fluxes during low tide were then calculated across the year according to these relationships against an entire year of actual emersion durations and irradiances. The use of a simulating model of the hourly fluxes frees from methodological constraints involved in upscaling up to annual fluxes. The results contribute to better evaluate the ecosystem functioning of these shores during emersion and are discussed in light of the consequences of canopy loss.

Materials and methods

Study site

The study site (48°43′48″ N, 3°59′12″ W) is an islet among an area of stacked boulder field and fractured bedrock in front of the Station Biologique de Roscoff, southwest of the English Channel (Brittany, France). This is a sheltered site characterised, from mean high water neap tide [6.3 m above Chart Datum (C.D.)] to mean low water neap tide (3.3 m above C.D.), by a dense cover of the fucoid Ascophyllum nodosum. The carbon budget was calculated for the mean height of the A. nodosum zone of the study site, which is 4.8 m above C.D.

Steps to the estimation of an annual carbon budget

The estimation of the annual carbon budget of the A. nodosum zone during emersion is based on a combination of: (1) direct measurements of CO2 fluxes newly acquired from April 2007 to June 2008 and previously obtained measurements from January 2006 to March 2007 (Golléty et al. 2008), which show how fluxes standardised for biomass vary as a function of the most influential abiotic factors, i.e. duration of emersion on a tidal scale, light on an hourly scale, and season on an annual scale; and (2) a modelling approach interpolating these flux variations as a function of the tide and light data for a case study year, 2008.

Abiotic factors

The calculation of the carbon budget is based on tide and irradiance data measured in 2008 near the study site. The emersion status of the site was defined every minute as either water height = 0 m (emersion) or water height >0 m (not emersion). For this purpose, the water height was calculated by linearly interpolating (approx function, R Development Core Team 2017) the tidal information measured every 10 min by the Roscoff observatory with respect to C.D. and then adjusted to the mean height of the study site.

The irradiance in the air is a dataset provided by the Service d’Observation du Milieu Littoral, INSU-CNRS, Station Biologique de Roscoff. Irradiance was measured as a 15-min average by a PAR (400–700 nm) sensor on the roof of the Station Biologique de Roscoff. To disentangle the general effect of irradiance from that of the cloud cover of 2008, a second light dataset assuming no cloud cover was created for every minute of the year. This theoretical irradiance was calculated each minute from the following sinusoidal curve (Lizon et al. 1998; Migné et al. 2004):

$$I\left(t\right)={I}_{{\text{m}}}\mathrm{ sin}\left(\frac{\pi t}{D}\right)$$
(1)

where Im (μmol m−2 s−1) is the maximal irradiance during the day, t (hours) is the time since sunrise and D (hours) is the daylength. Daylength was calculated as the number of hours between sunrise and sunset based on the civil twilight in Brest (Brittany, France) in 2008 (http://www.ephemeride.com/) Im was calculated from the theoretical greatest maximal daily irradiance (i.e. the zenith irradiance on the 21st of June) and the theoretical smallest maximal daily irradiance (i.e. the zenith irradiance on the 21st of December) measured at the study site and expressed using the following sinusoidal function:

$${I}_{{\text{m}}}={z}_{{\text{mean}}}-{z}_{{\text{ampli}}}{\text{cos}}\left(\frac{2\pi d}{366}+f\right)$$
(2)

where zmean is the mean between the theoretical maximum and the theoretical minimum zenith, zampli is the amplitude, d is the day of the year and f is a constant calculated so that the maximum light is reached on the 21st of June. In Roscoff, the greatest and the smallest maximum monthly irradiances measured in 2008 were recorded to be 2059 μmol m−2 s−1 in July and 717 μmol m−2 s−1 in December. In order for the theoretical irradiance to be always greater than the measured one, however, we applied a correction so that zmean = 1588 μmol m−2 s−1, zampli = 671 μmol m−2 s−1 and f = 0.172.

Response of CO2 fluxes to abiotic parameters

CO2 fluxes at the rock-air interface were measured based on the method described in Migné et al. (2002). Changes in CO2 mole fraction (ppm) in air upon rocky substratum were measured with an infrared CO2 gas analyser (LI-COR LI-800; LI-COR Inc., Lincoln, NE, USA). Using a data logger (LI-COR LI-1400; LI-COR Inc.) with a 15 s logging frequency (Fig. 1), CO2 fluxes were calculated from recorded data using the slope of CO2 concentration (µ\({\text{mol}}_{{{\text{CO}}_{2} }}\) molair−1) against time (min) and then expressed in carbon unit (mgC m−2 h−1) assuming a molar volume of 22.4 L mol−1 at standard temperature and pressure and a molar mass of 12 gC \({\text{mol}}_{{{\text{CO}}_{2} }}^{ - 1}\). A benthic chamber enclosed an area whose border was sealed using neutral silicon to obtain airtightness: the chamber comprised a square, transparent Plexiglas base, a gasket with an O-ring on each side and a transparent Plexiglas dome, thus enclosing 18.94 L and a 0.09 m2 rock area (Fig. 1). The chamber was placed on a relatively flat area enclosing natural densities of the canopy, lying flat on the rock surface. Self-shading resulting from placing the canopy inside the chamber was not higher than under natural conditions, as tested by light measurements taken inside the chamber above and inside the canopy. Unless specified otherwise, flux measurements were performed under maximal photosynthetically available radiation (PAR) at zenith of each month thus measuring community net primary productivity (NPP). Such measurements, including previously acquired data (Golléty et al. 2008), can thus be equated to maximal productivity values. Then, the chamber was opened to allow for equilibration with atmospheric CO2 concentration, which was reached in a few minutes. Finally, it was covered with an opaque polythene sheet in order to perform a second incubation under darkness, thus measuring community respiration (CR) assuming that CO2 increase measured in darkness after an abrupt transition from ambient light reflects the respiration activity of the community (Bordeyne et al 2017). Kremer (1980) has shown that photorespiration in Fucales, including Ascophyllum nodosum, is much lower than in green or red algae. Bidwell and McLachlan (1985) also showed that photorespiration is unlikely in A. nodosum at normal O2 and CO2 concentrations. Accordingly, incubations were sufficiently short (3–10 min) to avoid any changes in temperature and O2 or CO2 concentration to affect the metabolic rates within the chambers. Community gross primary productivity (GPP) was then calculated as the sum of NPP and CR.

Fig. 1
figure 1

CO2 fluxes measured on the Ascophyllum nodosum zone using a circuit of CO2 analysis and data logger (top left) connected to the squared-base benthic chamber with transparent dome (top right), together with the frame used to hold the neutral grey light filters (bottom left) that allows access to the chamber between measurements (bottom centre) and holds light filters between the chamber and the sunlight (bottom right)

Full size image

Variation of photosynthesis and respiration over the emersion period

CO2 flux measurements were performed in September 2007 and April and May 2008 to estimate the change in GPP and CR along the emersion period (e.g. effect of desiccation). At each date, NPP and CR were measured around zenith, with PAR always above saturation, each hour of the spring tide emersion on a single quadrat. Then, GPP or CR were expressed as the percentage of the initial GPP or CR, respectively, of each emersion (%GPPi and %CRi, respectively) and plotted as a function of time since the beginning of the emersion period. For %GPPi, the non-linear relationship was best fitted with a quadratic function:

$${\mathrm{\%GPP}}_{i}={a}_{1}{{t}_{E}}^{2}+{a}_{2}{t}_{E}+{a}_{3}$$
(3)

where tE is the time in min since the beginning of the emersion period and a1, a2, a3 are the parameters to be fitted. For %CRi, the data were fitted with a linear model:

$${\mathrm{\%CR}}_{i}={b}_{1}{t}_{E}+{b}_{2}$$
(4)

where b1 and b2 are the parameters to be fitted.

Response of photosynthesis to irradiance

From April 2007 to June 2008 (Table 1), 8 measurements of the relationship between benthic community net primary productivity and irradiance (P–I curves) were performed in situ combining the system described above with a series of neutral grey filters of varying intensity (Fig. 1).

Table 1 Photosynthetic parameters estimated on the Ascophyllum nodosum zone from April 2007 to June 2008
Full size table

Around the chamber, a PVC frame gave access to the chamber while the angled surface was oriented towards the sun and enabled maintaining different combinations of filters (Fig. 1). These were first used to measure CR under darkness, then NPP as a function of irradiance. The filters gave a mean ± S.D. (n = number of P–I curves with successful incubation for each filter combination) percentage of PAR reaching the benthic chamber of 0.0 ± 0.0% (n = 8), 1.0 ± 0.4% (n = 8), 3.4 ± 1.4% (n = 6), 6.9 ± 2.1% (n = 8), 13.9 ± 3.1% (n = 8), 27.2 ± 6.4% (n = 6), 49.4 ± 2.9% (n = 7) and 66.2 ± 11.7% (n = 8), as measured by a PAR Sensor (LI-COR LI-190) placed inside the frame on the side of the dome. For each measurement cycle, an additional incubation was performed without any filter. In the absence of any sign of photoinhibition, as expected for measurements performed at the community scale (Hubas and Davoult 2006; Tait and Schiel 2010; Bordeyne et al. 2017), these data were fitted to the model of Webb et al. (1974) to estimate the photosynthetic parameters (Pm and Ik):

$${\text{GPP = Pm}}\left[ {1 - \exp \left( {\frac{ - I}{{I_{k} }}} \right)} \right]$$
(5)

where Pm (mgC m−2 h−1) is the maximal GPP, I (μmol m−2 s−1) is the measured irradiance and Ik (μmol m−2 s−1) is the saturation onset parameter, defined as the irradiance at which Pm would be reached if GPP kept increasing linearly with irradiance. These measurements also allowed us to calculate the parameter of compensating irradiance Ic (μmol m−2 s−1), defined as the irradiance at which the magnitude of GPP equals that of CR.

Macroalgal biomasses

After any incubation measurements, the few non-encrusting macroalgae (Cladophora, Chondracanthus sp., F. serratus, Ulva intestinalis on the substrate, Fucus vesiculosus, Vertebrata lanosa, Elachista fucicola epiphytic on A. nodosum) were collected to estimate their biomass (Btot) present under the benthic chamber. V. lanosa was the most frequently found epiphyte, often contributing up to 99% of the epiphyte biomass except on the rare occasions when a small F. vesiculosus individual was present. Biomasses were estimated after drying for 48 h at 60 °C (expressed in kgDM m−2 where DM is dry mass). Except on the rare occasions when some F. vesiculosus was present, in which case it contributed up to 24% of the total biomass, algae other than A. nodosum always contributed under 10% of the total biomass. Each Pm and GPP value was standardised by the biomass present inside the chamber during that measurement (Pm/Btot and GPP/Btot).

To account for the actual macroalgal community biomasses present on the study site, a seasonal estimation of the canopy biomass (i.e. A. nodosum and its epiphytes, kgDM m−2) was performed. For this purpose, 8 to 12 50 × 50 cm quadrats were sampled along a transect perpendicular to the shore zonation in April (n = 12), June (n = 12) and November (n = 8) 2007 and January (n = 9) 2008. Canopy biomass was estimated using a size: weight allometric relationship as described in Golléty et al. (2011). The canopy biomass surveyed on the study site did not vary significantly seasonally [ANOVA, F(3,37) = 1.39, P = 0.261] and averaged 4.3 ± 3.5 kgDM m−2 over the year. The mean canopy biomass estimate was then used to estimate the monthly macroalgal biomass present on our study site (Bm, kgDM m−2) from the proportion that A. nodosum and its epiphytes represented monthly under the benthic chamber with respect to Btot during the monthly flux measurements (X ± SD = of 97 ± 4%, n = 12) (Golléty et al. 2008) and was expressed in gC m−2 (SI 1) considering a carbon content of 38% of the dry mass for macroalgae (Westlake 1963).

Seasonal variations of the metabolic parameters

It has already been shown, both for macroalgae (Golléty et al. 2008; Bordeyne et al. 2020) and microalgae (Migné et al. 2004), that variations of photosynthetic parameters along the year display a sinusoidal response due to both seasonal changes in day length and temperature. Sinusoidal functions were thus fitted both to the previously acquired CR data (updated fit compared to Golléty et al 2008 for the sake of methodological homogeneity, SI 2) and to the newly acquired maximal productivity data (P/Btot) and saturation onset parameter data (Ik) in order to estimate their variations along the year:

$$\varphi ={e}_{1}+{e}_{2}sin\left(\frac{2\pi d}{365}+{e}_{3}\right)$$
(6)

where φ is either P/Btot, Ik or CR, d is the day of the year and e1, e2, and e3 are the parameters to be fitted. In the case of the maximal productivity, the fitted data combined both the Pm/Btot data extracted from the photosynthesis to irradiance relationship (Eq. 5) together with the GPP/Btot data obtained from the monthly measurements of GPP (Golléty et al. 2008) to improve the strength of the response to light variations over an annual scale (SI 2).

Annual carbon budget

The carbon budget was calculated under emersion conditions only. Using either light data every 15 min (measured irradiance) or every minute (theoretical irradiances, Eq. 1), Pm/Btot was first calculated either every 15 min or every minute from daily Pm and Ik values (Eq. 6) combined with available measured or theoretical irradiance data, respectively. Ptot was then calculated as the product of Pm/Btot, the proportion of GPP since the beginning of emersion (%GPPi, Eq. 3) and Bm the monthly biomass estimated on our study site. The respiration carbon budget was calculated for every minute as the product of daily CR values (Eq. 6) and the proportion of CR since the beginning of emersion (%CRi, Eq. 4).

The annual community gross primary productivity (GPPa) and annual community respiration (CRa) under emersion were calculated as the sum of productivity or respiration across the year. Productivity under theoretical irradiance and respiration were summed for every minute. Productivity under measured irradiance was summed for every 15-min period. The annual community net primary productivity (NPPa) under emersion is the sum over the entire year of the difference between the community gross primary productivity and the community respiration calculated for each time step. For the budget based on measured irradiance, we used the community respiration values summed over each 15-min period. It has to be noted that these differences in calculation time-steps create artificial discrepancies at the annual scale when calculating separately the net productivity under light or darkness, such that their sum does not precisely equates NPPa.

Statistical analysis and computational tools

Differences in the seasonal estimates of canopy biomass of the study site were tested using an analysis of variance (ANOVA) checking for normality (Shapiro–Wilk test) and homoscedasticity (Bartlett test) (R Development Core Team 2019).

Model fitting (linear model or Nonlinear Least Squares, nls), and the calculation of the budget including the uncertainty and sensitivity analyses (see below) were performed using the R environment (R Development Core Team 2019) and all code is readily available (https://doi.org/10.5281/zenodo.10512341).

A model evaluation was performed following the recommendations from Owen-Smith (2007) consisting of an uncertainty analysis (either measured or theoretical light conditions) and a sensitivity analysis (measured light conditions only) of all the estimated parameters of Eqs. 3, 4 and 6. The sensitivity analysis was performed separately on the gross primary productivity budget and on the respiration budget. On the gross primary productivity budget, a comparison was performed on the relative importance of a ±10% change in (1) the speed at which GPP changes during emersion (parameter a2 of Eq. 3), (2) the average daily Pm/Btot value (parameter e1 of Eq. 6) and (3) the average daily Ik value (parameter e1 of Eq. 6). On the respiration budget, a comparison was performed on the relative importance of a ±10% change in (1) the speed at which CR changes during emersion (parameter b1 of Eq. 4) and (2) the average daily CR value (parameter e1 of Eq. 6).

We ran a parametric bootstrap of our model with 1000 replicates to estimate the uncertainty in the community carbon budget (daily, monthly and annual) due to all the parameter uncertainties. For each replicate, model parameters were drawn from independent normal distributions with means and standard errors estimated from the statistical fits for each parameter (Eqs. 3, 4, 5, and 6).

Results

Response of CO2 fluxes to abiotic parameters

GPP and CR values measured over a low tide period showed a different response of either process as a function of time since the beginning of the emersion period. At the onset of low tide, gross primary productivity initially increased for up to 1–2 h and then decreased until the end of the emersion period, while PAR remained saturating (September: from 1124 to 1779 μmolphotons m−2 s−1; April: from 1163 to 1995 μmolphotons m−2 s−1; May: from 1274 to 1840 μmolphotons m−2 s−1 (Fig. 2c). By the end of each emersion date, GPP represented 95, 64, and 57%, respectively, of the fluxes measured at the beginning of the emersion (Fig. 2a). Instead, community respiration decreased continuously during the emersion. By the end of the emersion, CR represented 52, 43, and 54%, respectively, of the fluxes measured at the beginning of the emersion (Fig. 2b). Parameters of fitted regressions (Eqs. 3 and 4) are given in Table 2.

Fig. 2
figure 2

Change in GPP (a) or CR (b) expressed as a percentage of the corresponding initial flux during the emersion period and average PAR during light incubations (c) from measurements performed in September 2007 (squares), April 2008 (circles) and May 2008 (triangles) and summarised by the fitted regressions (solid line) as %GPPi = −0.00104tE2 = 0.24550tE + 95.32015 and %CRi = −0.199tE + 113.380, respectively

Full size image
Table 2 Parameter estimates ± standard errors and associated R2 values given for each fitted regression used in the different steps of the model
Full size table

The high Ic values (Table 1) confirm that, during emersion, the magnitude of the community respiration fluxes represents a large proportion of the GPP fluxes. The fitted sinusoidal curves for P/Btot and Ik values (Eq. 6, Table 2) resulted in the following extreme values: the minimum P/Btot (59.5 mgC kgDM−1 h−1) was reached on the 20th of December and the maximum (138.6 mgC kgDM−1 h−1) on the 20th of June; the minimum Ik (23.5 μmolphotons m−2 s−1) was reached on the 3rd of February and the maximum (400.4 μmolphotons m−2 s−1) on the 4th of August.

The fitted sinusoidal curve for CR (Eq. 6) resulted in the following extreme values: the minimum CR (7.4 mgC m−2 h−1) was reached on January the 7th and the maximum (190.0 mgC m−2 h−1) on July the 8th.

Annual carbon budget

Under theoretical irradiances, gross primary productivity of the emerged community varied seasonally from 0.31 ± 0.11 gC m−2 d−1 (2nd January) to 5.42 ± 1.40 gC m−2 d−1 (13th June) at the daily scale (Fig. 3a) and from 33 ± 11 gC m−2 mo−1 (December) to 124 ± 33 gC m−2 mo−1 (July) at the monthly scale. GPPa was 859 ± 220 gC m−2 year−1.

Fig. 3
figure 3

Calculated carbon budget of the benthic community of the Ascophyllum nodosum zone during emersion showing a the daily gross primary productivity under measured (light grey) and theoretical (dark grey) irradiances, b the daily community respiration (black) and c the daily net primary productivity under measured (light grey) and theoretical (dark grey) irradiances

Full size image

Under measured irradiances, gross primary productivity of the emerged community varied seasonally from 0.03 ± 0.03 gC m−2 d−1 (3rd January) to 3.63 ± 1.00 gC m−2 d−1 (23rd July) at the daily scale (Fig. 3a) and from 19 ± 7 gC m−2 mo−1 (November) to 92 ± 26 gC m2 mo−1 (June) at the monthly scale. GPPa was 659 ± 176 gC m−2 yr−1. The sensitivity analysis of the emersion effect showed that a 10% change in the rate of speed of change of GPP with time since the beginning of emersion (parameter a2 from Eq. 3) resulted in a 4% change in GPPa. In comparison, a 10% change in average daily Pm/Btot and Ik value resulted in a 10% and 2.5% change in GPPa, respectively.

Emerged community respiration varied seasonally from 1.05 ± 0.29 gC m−2 d−1 (3rd January) to 4.42 ± 0.49 gC m−2 d−1 (30th June) at the daily scale (Fig. 3b) and from 38 ± 11 gC m−2 mo−1 (January) to 128 ± 15 gC m−2 mo−1 (July) at the monthly scale. CRa was 978 ± 105 gC m−2 year−1. The sensitivity analysis showed that a 10% change in the rate of speed of change of CR with time since the beginning of emersion (parameter b1 of Eq. 4) resulted in a 4% change in CRa. In comparison, a 10% change in average daily CR value (parameter e1 of Eq. 6) resulted in a 10% change in CRa.

Under theoretical irradiances, net primary productivity of the community varied seasonally from −1.64 ± 0.58 gC m−2 d−1 (23rd October) to 1.18 ± 1.30 gC m−2 d−1 (13th June) at the daily scale (Fig. 3c) and from −34 ± 26 gC m−2 mo−1 (August) to 3 ± 23 gC m−2 mo−1 (February) at the monthly scale. NPPa was −119 ± 244 gC m−2 yr−1 (292 ± 228 gC m−2 year−1 under light).

Under measured irradiances, net primary productivity of the community varied from −3.31 ± 0.73 gC m−2 d−1 (27th August) to 0.42 ± 0.51 gC m−2 d−1 (26th January) at the daily scale (Fig. 3c) and from −53 ± 20 gC m−2 mo−1 (August) to 0 ± 18 gC m−2 mo−1 (February) at the monthly scale. NPPa was −299 ± 204 gC m−2 year−1 (108 ± 185 gC m−2 year−1 under light, Fig. 4).

Fig. 4
figure 4

Calculated daily net primary productivity under emersion (thick line) and calculated confidence interval (thin line) estimated under measured irradiances and taking place only when light is available (NPP = Production—daylight respiration)

Full size image

Discussion

Improvement to previous CO2 fluxes estimations

A year-long study (Golléty et al. 2008), based on monthly measurements of GPP and CR under saturating irradiances during emersion, had previously been performed on the same Ascophyllum nodosum area. It estimated that the canopy was responsible for 92.6 and 94.6% of GPP and CR, respectively. It also showed that the benthic macrofauna community did not contribute significantly to the biomass nor to the CR fluxes (Golléty et al. 2008). In addition, it demonstrated that community emersion fluxes were fairly high compared to other estimates reported for laboratory studies of the same species, in spite of potential self-shading that might affect primary production. The in situ survey of other mid-intertidal communities dominated by other canopy-forming species confirmed that these macroalgal systems are responsible for high production fluxes during emersion (Bordeyne et al. 2015), even higher than during immersion (Bordeyne et al. 2017). The first estimation of the annual carbon budget of an intertidal rocky reef macroalgal community (Bordeyne et al. 2020) showed that it was autotrophic and that carbon was mainly assimilated during emersion periods. The present study builds upon previous measurements by integrating two sets of new measurements, namely (1) hourly flux measurements performed throughout emersion periods, and (2) hourly flux measurements as a function of irradiance performed throughout the year. The light saturation curves based on in situ measurements and considering seasonal variations were needed to refine saturating level estimations for the A. nodosum zone (i.e. the whole algal community). This approach improves our estimation for the maximum amount of carbon produced during a diurnal emersion. These results are also new evidence that there is absence of photoinhibition for the A. nodosum canopy. Finally, combining previous and new measurements in an analysis that integrates variations of production and respiration as a function of abiotic factors at different temporal scales allows estimation of daily, monthly and annual carbon fluxes during emersion, with a relatively low sensitivity to any of the parameters used. This approach highly improves the estimation of the carbon budget compared to simply upscaling hourly values taken at different seasons. The comparison between annual gross primary productivity budgets based on theoretical vs measured irradiance (GPPa 1.3 times higher under theoretical irradiances than under measured ones) further emphasises the importance of light in controlling carbon budgets at the annual scale. Unlike hypothesised, the study shows that, at the annual scale, the functioning of this ecosystem is heterotrophic during emersion. This unexpected result gives some insight into the functioning of such ecosystems under environmental changes.

Overall, the budget is not very sensitive to any of the parameters used. Both gross primary production budget and respiration budget were less sensitive to the effects of emersion duration (including desiccation) than to seasonal variations of abiotic factors. More specifically, it was the average daily CR and Pm/Btot values that were most sensitive, while the production budget was little affected by the seasonal variations in saturating irradiances. This suggests that future improvements of the budget should concentrate on factoring the seasonal variations of the effects of the different abiotic factors (such as temperature and nutrients) on either productivity or respiration (Bordeyne et al. 2015, 2020). In particular, additional measurements of both production and respiration as a function of temperature would help better understand the effect of this parameter, including the interplay between temperature and changes from emersion to immersion or the different strength of the temperature effect on either production and respiration (see below).

Confidence in the orders of magnitude is further provided by the sensitivity and uncertainty analyses. The former shows that the model is most sensitive to the parameters belonging to the individual adjustment with the lowest explanation power of the measured data (Table 1). This confirms the importance for future studies to improve the estimated effect of seasonal variations in abiotic factors, in particular that of light on gross primary productivity. The uncertainty analysis lends coefficients of variation varying from 10% for the respiration budget to 27% for the gross primary productivity budget under measured irradiance, indicating that the estimation of the parameters of individual adjustments (Table 2) has a limited impact on the estimation of the annual budgets.

Abiotic constraints to the carbon budget: integrating different temporal scales

The present study integrates the effect of abiotic factors constraining the carbon budget at different interfering temporal scales: emersion duration (which includes desiccation), daily and seasonal light variations, and seasonal changes in temperature. Although temperature was not used as a modelling parameter, its importance in explaining the sinusoidal response of the daily fluxes throughout the year and its regulating effect on maximal fluxes has been emphasised before (Golléty et al 2008).

Emersion duration

Emersion duration had different effects on productivity and respiration: throughout the year, GPP first increased at the beginning of emersion and then decreased, while CR only showed a decrease during emersion (Fig. 2). This differential response to the effect of emersion has also been observed for Fucus vesiculosus and F. serratus community fluxes (Bordeyne 2016). An increase in either the photosynthetic or the respiration rate of fucoids during the early stage of drying has long been reported and explained (Bell 1993 and references therein). While tendencies are similar between these canopies, several differences should be noted. Most noticeably, maximum photosynthesis is reached much later for A. nodosum (123 min) and is still by 105% of initial GPP by 200 min, while it reached 77 and 70% for F. serratus and F. vesiculosus communities, respectively, by then, lower on the shore. On the contrary, the speed of decrease is much more similar for the community respiration rates (that reached ~75% of initial rate by 200 min) (Bordeyne 2016). Our result thus not only reinforces the assumption that photosynthesis and respiration processes have a tight but different water dependency at the scale of the community. It also suggests different responses of the effect of emersion on photosynthesis across communities or tidal levels or both.

Light variation

Previous values of the saturation onset parameter (Ik) for A. nodosum measured in air were only available for individual thalli and varied between 153 µmol m−2 s−1, from steady-state light curves performed on individual portions of algae sampled at Spiddal, Ireland in Winter (Nitschke et al. 2012) to 200 µmol m−2 s−1, from CO2 fluxes measured on a thallus sampled in the Tay Estuary, Scotland, without any specifications as to the season (Johnston and Raven 1986). Ik values measured in our study at the community scale varied from 34 µmol m−2 s−1 in late January to 487 µmol m−2 s−1 in mid-June. This higher range results from both seasonal acclimation and layering causing self-shading (Stengel and Dring 1998). In winter, only upper layers receive low light levels and the seasonal acclimation lends low Ik values that approach individual ones (Kirk 2011). In summer, there is a seasonal acclimation to higher incident irradiances that involves a wider range of perceived irradiances by the different canopy layers (Tait and Schiel 2018) such that, at the community level, irradiances must be very high in order to saturate production. Due to differences in their thallus shape, A. nodosum canopies experience more layering and self-shading than the other mid-intertidal species Fucus vesiculosus and F. serratus for which ranges of Ik of, respectively, about 250–800 and 200–500 µmol m−2 s−1 were measured at the community scale (Bordeyne 2016). Extreme values for saturating onset parameter lagged by about 45 days compared to extreme values of the maximal primary productivity standardised by biomass (reached on solstice dates, as expected given the seasonality of environmental forcings and daylight duration, Bordeyne et al. 2020). This could either reflect its higher sensitivity to temperature, contrary to what is assumed for phytoplankton by Kirk (2011), or the additional sensitivity of maximal productivity to nutrient availability (Bordeyne et al. 2015).

Compensating irradiances (Ic) varied from 71 µmol m−2 s−1 in early winter to 370 µmol m−2 s−1 in late spring. In January, unlike other intertidal assemblages dominated by fucoid algae (Tait and Schiel 2018), the gross primary productivity did not compensate the community respiration under low light conditions. An Ic value higher than the saturating irradiance in spring potentially indicates that the respiration increases with temperature faster or stronger than the productivity does. This phenomenon has already been observed for macroalgae (Tait and Schiel 2013), diatoms (Hancke and Glud 2004) and seagrasses (Masini and Manning 1997).

Under theoretical irradiances, the gross and net primary productivity budgets exhibit obvious fortnightly fluctuations (Fig. 3a and c). This pattern is characteristic of intertidal communities (e.g. Serôdio and Catarino 1999; Migné et al. 2004), where the diel light cycle interferes with the tidal cycle. Along the fortnightly alternation of spring and neap tides, the phasing between the tidal and the diel light cycle either maximises or minimises the irradiance available for primary productivity (SI 3). In the present study site, low water spring tides fall around midday, which maximises the available irradiance during emersion when the day length is short. As a consequence, the daily variability is minimal during equinox periods (Fig. 3c), when the highest tide amplitude falls during daylight hours and consecutive night-time emersions have similar durations. On the contrary, when the day length is long (around summer solstice), the irradiance available under emersion periods is maximised at neap tide, when two low tides occur under daylight. As a consequence, daily variability is maximal (Fig. 3c). These fortnightly variations are less obvious under measured irradiances: the effect of clouds on the available irradiance masks the effect of the superimposition of the tidal and diel cycles on the primary productivity. It is thus only under theoretical irradiances that the effect of these two superimposed cycles can be seen on the extreme values of net primary productivity, with maximum autotrophy in late spring and maximum heterotrophy in autumn (Fig. 3c).

Re-evaluation of the ecosystem functioning during emersion

Measurements performed in situ at the community scale allowed us to integrate interactions that are otherwise absent at the individual scale. This gives access to aspects of the functioning that cannot be otherwise identified. Integrating this response at the annual scale, through abiotic constraints, shows that the functioning of the ecosystem is heterotrophic during emersion. This result highlights the importance of community measurements in order to disentangle the functioning of the ecosystem.

High magnitude of ecosystem metabolism

Previous measurements of monthly variations of the hourly GPP and CR fluxes due to the community had already revealed the fairly high magnitude of both productivity and respiration fluxes of the community, under the dominating contribution of its canopy (Ascophyllum nodosum together with its epiphytes) (Golléty et al. 2008).

Daily estimates obtained here, based solely on emersion periods, still appear to be high compared to those from other studies that integrate periods of emersion and immersion over a 24-h period (Barrón et al. 2003; Attard et al. 2019). When standardised by biomass, our maximal daily estimates of both production and respiration are comparable to those estimated for a Fucus serratus community in Norway over an entire tidal cycle (Barrón et al 2003). When standardised by aerial surface, however, daily respiration rates estimated during emersion varied between 1.1 and 4.4 gC m−2 for the A. nodosum community while varying between 0.6 and 2.6 gC m−2 for a Fucus vesiculosus community and between 0.6 and 4.5 gC m−2 for the F. serratus community from the same location as our study site (Bordeyne 2016). Those estimates led to an annual budget of respiration higher for the A. nodosum community under emersion only than for either Fucus communities over the entire cycle (978, 587 and 875 gC m−2, respectively).

Heterotrophic budget during emersion

While hourly measurements under saturating light always resulted in a positive net productivity (Golléty et al. 2008), the combined effect of (1) high respiration values throughout the year, which lend compensation at high irradiances, sometimes higher than the saturating ones and eventually absence of compensation, and (2) irradiances rarely above saturation (16–45% of emersed duration each month) resulted in net heterotrophy measured here during emersion (NPP reaching −3.3 gC m−2 d−1). As discussed previously (Golléty et al 2008), the high respiration fluxes are likely the result of a higher EPS production during emersion (Carlson and Carlson 1984), subsidising the microscopic epiphytes (Khailov and Burlakova 1969) covering the A. nodosum fronds (Sieburth and Tootle 1981).

Community heterotrophy at the daily scale thus emerges in spite of high community primary productivity during emersion, under the combined effect of high community respiration and the amount of light available, being more limited during neap tides and especially so on cloudy days (SI 4). The effect of light availability is reinforced by the combined effect of the tidal and seasonal cycles as seen when extracting the budget during daylight hours only: during daylight emersions (Fig. 4), the budget is usually autotrophic during spring tides, especially those of winter and spring and mostly heterotrophic during neap tides, when emersions are shortest, and during autumn, when cloudiness is higher. At the monthly scale, heterotrophy is highest at the end of summer, when the community respiration is at its maximum while gross primary productivity suddenly drops, likely due to a nutrient limitation in the area (Bordeyne et al. 2015).

While autotrophic at the annual scale, the carbon budget established on the nearby Fucus communities also shows daily heterotrophy, but NPP during emersion never exceeded −0.6 gC m−2 d−1 (Bordeyne 2016). The high respiration fluxes measured on the F. serratus communities were thus largely compensated for by high production fluxes at the annual scale (Bordeyne et al 2020), most of it taking place during emersion. The A. nodosum community is under emersion for over twice as long as the F. serratus one, yet the higher respiration is not compensated for by higher production fluxes during those emersions. Additional comparisons between these communities would thus be needed to tease apart the contribution of abiotic factors, including emersion duration, to these differences in heterotrophic magnitude throughout the year.

Implications for the full carbon budget

The evaluation of the carbon budget during emersion provides only a partial understanding of the carbon balance of intertidal rocky reef macroalgal-dominated ecosystems. While additional measurements would have been needed to estimate the full carbon budget (Johnson 2022), various hypotheses of the effect of immersion on carbon fluxes can still be discussed to indicate gaps in the literature that need investigating.

Estimating the budget during immersion

Evaluating carbon flux variations during immersion, which represents 57.9% of the year on that site, involves evaluating both changes in: (1) community light-use efficiency (photosynthetic parameters), as a result of the change in 3-D structure of the canopy, as a function of both water height and light attenuation throughout the seasonal cycle; and (2) community respiration due to different microscopic heterotrophic activity on rocks and mainly on macroalgal surfaces notably with temperature changes between air and water.

Two hypotheses can thus be proposed for the functioning of the A. nodosum community. First, annual net primary productivity under immersion might compensate for the high annual heterotrophic budget estimated here under emersion at least during part of the year. This could result from two, non-mutually exclusive, mechanisms: it could be that there are lower respiration rates under immersion than under emersion, to a higher extent than that observed for Fucus serratus further down the same shore (Bordeyne et al. 2017), due to the important shedding during immersion of the heterotrophic biofilm covering A. nodosum fronds (Halat et al. 2015). It could also be that primary production is more efficient under immersion, notably at the beginning of immersion when fronds are raised with their ends floating on the water surface. This is particularly challenging to evaluate as the rising tide does not result in a sudden change from emersion to immersion conditions. In a modelling prospect, transition from emersion to immersion could be seen as a two-step process, depending on whether the water height is lower or higher than the maximum algal length. Using a corrected frond height would make for a more realistic model. It is expected that light use is optimised due to light flecking delivery (Dromgoole 1988; Wing and Patterson 1993), likely as a result of the arrangement and movements of the fronds in water as already suggested by Tait et al. (2014). The effect of the change in canopy structure, however, on light-use efficiency needs to be checked by direct in situ P–I measurements of entire communities under immersion (see for ex: Tait et al. 2014). A better use of the lower irradiances available during immersion would increase the primary productivity during immersion. It could well be that these responses are also a function of the adaptation of species to different tidal heights (Migné et al. 2015b). During emersion, the lower saturating irradiance measured for A. nodosum than for Fucus serratus or F. vesiculosus (Bordeyne 2016) would suggest higher productivities under this condition. At the daily scale, however, gross primary productivity is unlikely to be as high as those measured during emersion and, during autumn and winter NPP might still be heterotrophic (Bordeyne et al. 2017).

Second, the community might be heterotrophic at the annual scale even when considering the full tidal cycle due to the decomposition of autochthonous or allochthonous organic matter trapped under the canopy that could be responsible for high bacterial respiration. Indeed, compared to Fucus spp. canopies, the A. nodosum one traps abundant autochthonous organic matter under its longer fronds (pers. obs). This is probably reinforced by the longer emersion periods taking place at this shore level (42.1% of the time against 17.3% on the F. serratus canopy of the same site, Bordeyne et al 2020). In addition, community respiration is positively correlated with temperature both in air and water (Tait and Schiel 2013; Bordeyne et al. 2017). For a given temperature, however, it is expected to be lower under water (Bordeyne et al. 2017). The direction of change in hourly carbon fluxes from emersion to immersion will then depend on the time of year (Madsen and Maberly 1990) but might still result in a heterotrophic carbon balance at the yearly scale, even if periods of the year keep being autotrophic (Attard et al. 2019; Bordeyne et al. 2020).

Consequences of canopy loss

The understanding of the carbon budget of an intertidal rocky reef macroalgal community contributes to identifying the consequences of direct and indirect human activities on the functioning of these ecosystems. Locally, A. nodosum canopies can be impacted predominantly by harvesting, which will have short- and long-term consequences on ecosystem functioning. Experimental removal of F. serratus canopy results in a short-term reduction in primary productivity and respiration (Valdivia et al. 2012; Migné et al. 2015a) and temporal variation in community respiration (Valdivia et al. 2012). In addition, harvesting also has short-term effects on the contribution of A. nodosum to the local detrital pool (Halat et al. 2015) and long-term effects on the composition of the macroalgal stand, lending to the partial replacement of A. nodosum by Fucus spp., and the associated invertebrate diversity (Jenkins et al. 2004). Changes in the distributional limits of A. nodosum (Khan et al. 2018) could further reduce southern populations and have the same effects on the carbon budget as harvesting. The removal or reduction of A. nodosum canopy from sheltered rocky shores of North Atlantic will significantly alter the carbon budget and functioning of these areas. Interactions with consumers will likely determine whether these changes will result in increased autotrophy if the A. nodosum canopy is quickly replaced by a F. vesiculosus and/or F. serratus community (Bertness et al. 2002; Jenkins et al. 2004), or increased heterotrophy if on the contrary the canopy recovery is prevented by recruitment of sessile invertebrates (Bertness et al. 2002).

Conclusion

Our study shows the importance of integrating in situ measurements with variations of abiotic factors at the daily scale in order to identify the parameters responsible for the carbon balance between autotrophy and heterotrophy. The role of the abiotic factors that drive the carbon budget of Ascophyllum nodosum communities stresses the importance of gaining the best understanding of the functioning of these habitats in order to evaluate the consequences of the effects of global changes and make informed decisions on ecosystem conservation.