Carbon sequestration and geochronology in Southern England's seagrass meadows

Study sites

Carbon sequestration rates were calculated from mixed, intertidal, seagrass meadows incorporating Zostera marina (Eelgrass) and Zostera noltii species, located in four study sites along the Solent coast of southern central England [Figure 1]: Creek Rythe (CRST) in Chichester Harbour; Hayling Island (LGST) and Farlington Marshes (FMST) in Langstone Harbour; and Porchester (PMST) in Portsmouth Harbour. Portsmouth, Langstone, and Chichester harbours are low-energy coastal environments, with a micro/meso-tidal range forming natural estuaries and harbours^[36,37]. This region has some of the most complex tidal regimes in the U.K., caused by its natural coastal configuration and the proximity of a semi-diurnal tidal node in conjunction with historical natural and anthropogenically driven changes in water depth and coastal morphology, which has led to distortion of the tidal curve^[38]. As a result, patterns of sediment transport in the region have been described as more complicated than other coastal regions along the UK, with fluvial and cliff erosion inputs, and transport driven by tides, wave activity, and littoral drift, generally moving from east to west^[39-41] [Figure 1].

Figure 1. Map of the United Kingdom including a zoomed image showing the location of seagrass study sites in the Solent region (yellow), from west to east: Porchester (PMST), (Farlington Marshes (FMST), Hayling Island (LGST), and Creek Rythe (CRST). Coloured arrows show the patterns of sediment input and transport according to SCOPAC (STS)^[40,41]. Maps are adapted from Esri ArcGIS online base maps, including scale bars in Km.

Field methods

During the summer of 2017 (June-August), two 50cm deep, 7.5 cm diameter PVC corers were used to extract sediment cores from within intertidal seagrass meadows at each of the four study sites, resulting in a total of eight cores (labelled CRST1, CRST 2, FMST1, FMST2, LGST1, LGST2, PMST1 and PMST 2). This chosen core depth was expected to cover the last 150 years of sediment accretion, which is the maximum detection limit for ²¹⁰Pb dating^[42,43]. Sediment cores were carefully inserted into the sediment to allow minimal compression (< 10%)^[25] and then carefully removed by digging around and capping at the bottom, to be immediately packed, labelled, and sealed with industrial cling film and duct tape, to prevent remobilisation, degradation, and contamination.

Laboratory methods

All cores were stored in freezers at -26 °C before preparation for laboratory analyses. Each frozen core barrel was extruded from the PVC tubes by allowing the outer core surface to thaw and then forcing the frozen sediment out. Sediment compaction was then re-assessed by measuring the length of the core section before and immediately after extrusion and no compaction was found^[25]. Each sediment core was cleaned, logged, and sliced into 1 cm depth increment subsamples, to be analysed for ²¹⁰Pb, organic matter (converted to total organic carbon), sediment particle size, and soil bulk density^[44,45].

Each 1 cm subsample was oven-dried at 40 °C prior to gamma spectrometry analysis, until a constant weight was achieved, to determine soil moisture content and dry bulk density^[25]. Dried subsamples were then prepared for analysis by gently disaggregating the material using a pestle and mortar. Approximately 5-7 g of dried sediment from each subsample was carefully weighed into cylindrical plastic vials for determination of ²¹⁰Pb_total down core activities via gamma spectrometry^[25]. The remaining dried sediment from each sample was separated to be analysed for organic matter content, to calculate carbon sequestration rates and particle size.

Climate data

Historical climate data for the Solent region was obtained from the MET office archives, covering the period between 1900 and 2017 for daily maximum and minimum temperatures, wind speed, and precipitation. Daily values of maximum and minimum temperature, wind speed and precipitation were used to calculate a mean for each year of interest. Monthly sea level data were collected from the Permanent Service for Mean Sea Level (PSMSL) database for the Portsmouth station; data were available between 1955 and 2017. Monthly sea level values were used to calculate a yearly mean. The relationships between minimum and maximum temperature, wind speed, precipitation, sea level, and sediment C_seq rates (g m^-2 year^-1) calculated using both the CRS and conventional methods were then analysed.

Statistical analyses

Spearman’s Rho Correlation tests were used, due to the non-normal distribution of data, to assess the relationship between estimated C_seq rates (g m^-2 year^-1) and the following variables: %mud, D50 (µm), sorting coefficient (φ), minimum, and maximum yearly average temperatures (°C), precipitation (mm), average wind speed (kn), and sea level (mm).

Radionuclide dating

Two cores, LGST 1 and PMST 2, had ²¹⁰Pb excess activity profiles indicating the occurrence of mixing or bioturbation and were, therefore, not used to calculate C_seq rates using the CRS model [Figures 2]. While both Creek Rythe cores had a comparable count value supporting ²¹⁰Pb equilibrium reached at 11.28 Bq/Kg for CRST1 and 11.61 Bq/Kg at CRST 2 [Figure 2A and B], core CRST 1 presented the highest surface count values of ²¹⁰Pb excess, of 128.76 Bq/Kg [Figure 2A]. The cores from Farlington Marshes provided the lowest count values of supported ²¹⁰Pb at equilibrium, which were 7.84 and 7.24 Bq/Kg for FMST 1 and FMST 2, respectively [Figure 2C and D]. The lowest surface ²¹⁰Pb excess count was similarly found in the FMST1 core, at 32.39 Bq/Kg [Figure 2C]. The LGST cores had the highest maximum depth for ²¹⁰Pb excess activity detection (45 cm) [Figure 2E and F], and PMST 1 had the highest count values of supported ²¹⁰Pb at an equilibrium of 18.20 [Figures 2G].

Figure 2. (A-D) Total activity down-core for ²¹⁰Pb (A) CRST 1, (B) CRST 2, (C) FMST 1, and (D) FMST 2 represented by blue lines (diamond markers), with error calculations (5%) shown in red. Supported ²¹⁰Pb values represented by orange (square markers) lines on all graphs, on the left. Respective graphs on the right showing the Natural Log (Ln) of ²¹⁰Pb excess per depth (CF:CS model) with regression equations. (E-H) Total activity down-core for ²¹⁰Pb (E) LGST 1, (F) LGST 2, (G) PMST 1, and (H) PMST 2 represented by blue lines (diamond markers), with error calculations (5%) shown in red. Supported ²¹⁰Pb values represented by orange (square markers) lines on all graphs, on the left. Respective graphs on the right show the Natural Log (Ln) of ²¹⁰Pb excess per depth (CF:CS model) with regression equations.

Sediment accretion rates

The CF:CS model calculated regression between sediment accretion rates and depth revealed that the model was a good and statistical fit to the data from all cores (P < 0.05), with R² values ranging from 0.53 (PMST 2) to 0.94 (PMST 1) [Figure 2A-H]. The ²¹⁰Pb profiles' log-linear nature across all cores suggests steady accretion. According to the CF:CS model, the mean sediment accretion rates ranged from 1.2 mm year^-1 (CRST 1 and FMST 1) and 10.8 mm year^-1 (LGST 2) [Table 1]. For the CRST 1 and FMST 1 cores, the mean sediment accretion rates obtained by the CRS model were identical to those found using the simple CF:CS model, but higher than the CF:CS model for the PMST 2 core [Table 1]. Compared to CF:CS, all other cores’ average sediment accretion rates calculated by the CRS method were lower [Table 1]. Similar to the CF:CS technique, the LGST 2 core had the highest sediment accretion rate of 6.3 mm year^-1, and the CRST 1 and FMST 1 cores CRS had the lowest rates of 1.2 mm year^-1 [Table 1].

Carbon sequestration rates

Down-core profiles in %C_org displayed variable trends [Figure 3A-D], with some sites showing general declines in %C_org with depth, such as Creek Rythe and Porchester. For all sites, apart from Cowes, down-core distribution in %C_org was not monotonic, showing alternative increase and decrease with depth [Figure 3A-D]. Therefore, since %C_org calculation methods used in this study do not differentiate between buried and remineralised carbon, the uppermost 15 cm layer from each core was not considered for the calculation of C_seq rates using the CRS method. The mean %C_org calculated for all cores was 1.73 ± 0.46 (%), with cores from Farlington Marshes presenting the highest average downcore %C_org, 2.30 ± 0.34 (%) in FMST 1 and 2.0 ± 0.25 (%) in FMST 2, while cores from Porchester presented the lowest downcore %C_org values, 1.49 ± 0.20 (%) in PMST 1 and 1.36 ± 0.29 (%) in PMST 2 [Figure 3].

Figure 3. Down-core profile of %C_org for (A) CRST, (B) LGST, (C) PMST, and (D) FMST sampling sites.

C_seq rates were determined both by the conservative and ²¹⁰Pb CRS methods for all cores, apart from LGST 1 and PMST 2. For the conservative method, the lowest %C_org measured from each core was determined as follows: CRST 1, 0.88 (%); CRST 2, 1.43 (%); FMST 1, 1.84 (%), FMST 2, 1.58 (%); LGST 2 0.35 (%), PMST 1, 1.05 (%). Calculated C_seq rates by this method were remarkedly lower than the ones calculated by the CRS model for all cores [Table 2]. The average C_seq rate determined by using the conservative method was 15.38 ± 7.17 (g m^-2 year^-1) [Table 2].

For all cores, the average C_seq rate determined by using the ²¹⁰Pb CRS method was 67.91 ± 32.39 (g m^-2 year^-1) [Table 2]. In one of the cores from Hayling Island (LGST 2), the C_seq rates were highest at 106.05 ± 49.09 (g m^-2 year^-1), and in Farlington Marshes (FMST 1), they were lowest at 19.91 ± 10.72 (g m^-2 year^-1)[Table 2]. All cores presented sediments predominately muddy, fine and poorly sorted particles, with an average mud percentage (silt + clay) between cores of 86.80% ± 9.89%, a sorting coefficient of 1.97 ± 0.32 and a median particle size of 22.73 ± 7.96 (µm) [Table 2].

To assess the link between temperature and C_seq rates calculated by the CRS method, available daily temperature data from 1,900 were compiled at 12-h intervals (09:00 am-09:00 pm). In the analysed time period, the recorded annual maximum and lowest temperatures averaged 14.83 ± 0.23 (°C) and 7.53 ± 0.25 (°C), respectively [Table 2]. The analysis also included precipitation level data, with yearly averages of 2.18 ± 0.18 (mm) per day across all sites and collected during the same 12-h period each day [Table 2]. Daily maximum wind speed measurements were taken over the whole 24-h period, with an average of 9.69 ± 0.44 (kn) [Table 2]. The average reported sea level for the years under study was 7,074.01 ± 19.98 (mm) [Table 2].

In contrast to the CRST 2 core, where the oldest sediments were set down around 1947 [Figure 4B], the oldest sediments assessed for the CRST 1 core were laid down around 1906 [Figure 4A]. The oldest sediments in the FMST 1 core were deposited in approximately 1950 [Figure 4C], whereas the sediments in the FMST 2 core were older dated and had been deposited around 1939 [Figure 4D]. The dateable section in the core from the Porchester site was substantially younger, with the PMST 1 core having a maximum age of 1972 [Figure 4E]. The oldest section was 1921, found in the LGST 2 core [Figure 4F]. These were all revealed by ²¹⁰Pb dating using the CRS method.

Figure 4. CRS calculated the age of each depth for the (A) CRST 1, (B) CRST 2 cores for the Creek Rythe site, (C) FMST 1, (D) FMST 2 cores for the Farlington Marshes site, (E) PMST 1, for the Porchester site and (F) LGST 2 for the Hayling Island site.

For all sediment cores, the %mud increased through time [Figures 5A, 6A, 7A, 8A]. The lowest % of mud was recorded at the surface of the FMST 2, rising from 57.35% in 2017 to 63.79% about 1939 [Figure 6A]. In comparison to all other cores, the %mud for the LGST 2 core showed a considerable variance with age, moving from 77.78% in 2017 to the maximum value of 94.25% for this core than all other cores, changing around both 1994 and 1999 [Figure 7A]. Conversely, the %mud for the PMST 2 core had little variance throughout time, ranging from 94.70% around 2017 to 93.76% around 1954 [Figure 8A]. Although more recent sediments were generally better sorted than older ones for all cores, all sediment layers were classed as being poorly or very poorly sorted [Figures 5B, 6B, 7B, 8B]. For the Hayling Island (LGST 1 and 2) and Porchester (PMST 1 and 2) cores, the degree of sorting followed the same patterns of change with time as %mud [Figure 7B].

Figure 5. Distribution of % mud (A), Degree of sorting (B), ²¹⁰Pb CRS method derived sediment accretion rates (C), and Carbon sequestration rates (D), with age, for the CRST 1 and 2 cores.

Figure 6. Distribution of % mud (A), Degree of sorting (B), ²¹⁰Pb CRS method derived sediment accretion rates (C), and Carbon sequestration rates (D) with age, for the FMST 1 and 2 cores.

Figure 7. Distribution of % mud (A) and degree of sorting, (B) for the LGST 1 and 2 cores, ²¹⁰Pb CRS method derived sediment accretion rates (C), and Carbon sequestration rates (D) with age, for the LGST 2 core.

Figure 8. Distribution of % mud (A), Degree of sorting (B) for the PMST 1 and 2 cores), ²¹⁰Pb CRS method derived sediment accretion rates (C), and Carbon sequestration rates (D) with age for the PMST 1 core.

Sediment accretion rates ranged from 0.5 to 2.6 mm year^-1, between 1906 and 2016, for the CRST 1 core, peaking at 1906 (2.6 mm year^-1) and 1975 (1.2 mm year^-1) [Figure 5C]. While sediment accretion rates for the CRST 2 core ranged from 0.76 to 1.99 mm year^-1, between 1947 and 2016, peaking in 1982 (12.1 mm year^-1), and 1980 (4.9 mm year^-1) [Figure 5C]. The Farlington Marshes FMST 1 core had the highest accretion rates reported around 1984 (2.22 mm year^-1), ranging from 0.43 to 1.37 mm year^-1 between 1950 and 2017 [Figure 6C]. Moreover, sediment accretion rates for the FMST 2 core indicated an overall trend of rise through time, with a standstill during the 1950s and 1990s and greater accretion rates reported around 2017 (2.21 mm year^-1) and 2007 (1.65 mm year^-1) [Figure 6C]. In general, sediment accretion rates decreased with age for the LGST 2 core, ranging from 0.77 to 6.50 mm year^-1, between 1921 and 2017 [Figure 7C]. For the Porchester, PMST 1 core, accretion rates varied between 0.82 to 2.37 mm year^-1, between 1972 and 2017 [Figure 8C].

The greatest CRS calculated C_seq rate for the Creek Rythe, CRST 1 core, was recorded in 1930 at 35.68 g m^-2 year^-1, while the lowest was recorded in 2016 at 10.3 g m^-2 year^-1 [Figure 5D]. Comparatively, the CRST 2 core showed higher C_seq rates, with the highest value occurring in 1982 at 296.5 g m^-2 year^-1, and the lowest in 1947 at 18.6 g m^-2 year^-1 [Figure 5D]. The Farlington Marshes FMST 1 core’s CRS calculated C_seq rates ranged from 7.31 in 1950 to 35.38 g m^-2 year^-1 in the years surrounding 1984, while C_seq rates for the FMST 2 core showed maximum values of 81.03 g m^-2 year^-1, presented around 1958, and the lowest of 12.36 g m^-2 year^-1 shown around 1939, demonstrating substantial oscillations over the dated time period and not following the same pattern as sediment accretion rates in this core [Figure 6D]. The LGST 2 core recorded the highest values around 1956, at 184.80 g m^-2 year^-1, and the lowest values around 1921, at 15.57 g m^-2 year^-1, showing contrasting patterns to sediment accretion rates [Figure 7D]. For the Porchester PMST 1 core, in addition to declining with age, CRS calculated C_seq rates have been declining more recently than sediment accretion rates, up until 2010, peaking in 2017 at 246.08 g m^-2 year^-1 and reaching their lowest level in 1972 at 22.02 g m^-2 year^-1 [Figure 8D].

Relationships between environmental variables and carbon sequestration rates

C_seq rates calculated by both the CRS [Table 3] and conventional methods [Table 4] demonstrated no significant relationships for most sampling sites). Precipitation levels showed a negative effect on carbon sequestration processes calculated by the conventional method in Creek Rythe’s cores (CRST), as can be seen from the significant negative relationship between the two variables (r_s = -0.757; P < 0.05) [Table 4].

Radiometric dating

This study estimated sediment accretion rates from seagrass sediment cores in the Solent region, U.K. using ²¹⁰Pb activity. The similar accretion rates seen across all cores when measured by the CF:CS and CRS methods provide evidence that these methods apply to the study area. However, there were variations in the maximum depth of ²¹⁰Pb_excess detection, sediment accretion rates, and C_seq rates between cores from the same study sites when calculated using the CRS method. Results also showed large differences between C_seq rates calculated by the CRS and the conventional approach, which indicate the need for further direct and local measurements of C_seq rates, using reliable methods and mathematical models, rather than using global or regional estimations for blue carbon budgets.

The use of ²¹⁰Pb to assess sediment accretion rates

The CRS method utilised in this study has been widely applied in estuarine and vegetated coastal environments due to its robustness against non-monotonic features in the ²¹⁰Pb record and is assumed to be relatively insensitive to mixing^{[43,53,56,57]}. The model assumes a constant direct ²¹⁰Pb atmospheric fallout, combined with fluctuations in the sedimentation rate and variable initial specific activity, calculated in combination with sediment accumulation rates, which may occur in response to natural processes or anthropogenic influences^[58]. Moreover, the CF:CS model is also, to some degree, able to cope with temporal variations in mass accumulation rates, such as mixing at the surface layers^[14]. Arias-Ortiz et al. listed seven distinct types of ²¹⁰Pb_excess specific activity profiles that can be identified in vegetated coastal sediments based on a review of the literature^[14]. These types have been classified in terms of “ideal” profiles, produced by constant sediment accumulation in steady-state conditions, and ones that would show one or more of the most common disturbances encountered in vegetated coastal sediments^[14]. The most common disturbances listed are the presence of mixing (physical or bioturbation), increasing sediment accumulation rates, erosion, or alteration by intrinsic features of sediments such as heterogeneous grain size, distribution, and decay of organic matter. Based on this classification, the sediment cores from Hayling Island (LGST 1 and 2) and Creek Rythe (CRST 1 and 2) showed ²¹⁰Pb_excess specific activities that possibly correspond to mixing attributed to higher rates of sediment accretion^[59-62], and mixing as a result of bioturbation or sediment resuspension (CRST 1, LGST 1)^{[14,21,35,63,64]}. More specifically, the CRST 2 core displayed a ²¹⁰Pb_excess activity profile consistent with the deposition of allochthonous older material^[65]. Moreover, the Porchester PMST 1 core showed a scattered ²¹⁰Pb_excess activity profile, which might reflect the periodic occurrence of mixing, resuspension, and allochthonous input, causing repetitive reworking in the overall mixed sediment column^[21,66]. Most cores presented ²¹⁰Pb_excess activity profiles consistent with erosion processes, except the CRST 2 core from Creek Rythe and LGST 2 core from Hayling Island^[13,21,67]. Ideally, additional cores would be needed to identify possible patterns of erosion, mixing, or disturbance on the sediment profiles from the sampled sites. However, given the large agreement in accretion rates calculated by both methods and the good fit to the data presented by the CF:CS method, ²¹⁰Pb activity profiles have been deemed a worthy method to provide this initial assessment of sediment accretion rates for the sampled sites.

Sediment accretion and C_seq rates

Sediment accretion rates calculated for most sites, by both the CRS and CF:CS methods, ranged from 0.61-6 mm year^-1, agreed with the reported global range of 2.02 ± 0.44 mm year^-1[68]. However, Hayling Island’s LGST 2 core was the only exception, producing higher average accretion rates of 6.3 mm year^-1, using the CRS method, and 10.8 mm year^-1, using the CF:CS method. Moreover, this study’s mean sediment accretion rates were generally higher than those reported for Western Australia’s Posidonia australis meadows by Marba et al.^[13], of 0.67 ± 0.03 mm year^-1, and for temperate, tropical and sub-tropical multi-species seagrass meadows along East and Southeast Asia, by Miyajima et al.,^[20] of between 0.32-1.34 mm year^-1. Most study sites, apart from Hayling Island, had cores with calculated accretion rates similar to those reported by Jankowska et al.^[62] (1.3 ± 0.2 mm year^-1) and Poppe and Rybczyk^[3] (1.9 ± 0.1 mm year^-1) for Z. marina meadows. However, Porchester and Creek Rythe also presented cores with higher mean sediment accretion rates, comparable to Mediterranean Posidonia oceanica meadows, of between 2.4-4.2 mm year^-1[19], demonstrating the variability in sediment accretion within sites.

The measured supported ²¹⁰Pb ranged from 7.24 Bq kg^-1 in Farlington marshes (FMST 2 core) to 18.20 Bq kg^-1 in Porchester (PMST 1). Not too different from the ones described by a previous study, analysing salt marshes within the Solent region, of 12.2 Bq kg^-1 supported ²¹⁰Pb on average^[38]. Sites presenting low sediment deposition would show a ²¹⁰Pb inventory lower than expected, and bioturbation would be responsible for the apparent long-term accretion rate, while a higher-than-expected inventory could imply that the site is depositional, and the apparent long-term accretion rate reflects both bioturbation and accretion^[3,59,69,70]. Results from this study support these assumptions, with higher sediment accretion rates found in Porchester (PMST 1), of 4.2- and 4.3-mm year^-1, than Farlington Marshes (FMST 2), of 1.6- and 1.2-mm year^-1, when calculated by both the CRS and CF:CS methods, respectively. This is because, despite the input of fluvial sediment in both sites, Porchester is less exposed to tidally driven sediment mixing than Farlington Marshes, which could explain the lower sediment accretion rates, possibly caused by sediment flushing [Figure 1]^[41].

Patterns of %C_org profiles for all sediment cores from this study were non-monotonic, potentially representing the surface mixed layer where bioturbation processes often occur^[55]. However, it is important to highlight that even though the approach taken by the CRS method used in the study does not take erosion/bioturbation and remineralisation into account, and there is no differentiation between autochthonous and allochthonous carbon, the study aimed to provide an initial estimation of C_seq processes in the study region^[55]. Mean C_seq rates, calculated by the CRS method, from seagrass meadows in the studied sites, of 75.12 g m^-2 year^-1, were comparable to other global regions^[67]. Furthermore, a review of seven studies summarising a total of 123 sites, reported an estimated range of C_seq rates from 45 to 190 g m^-2 year^-1, even though the mean calculated rate of 138 ± 38 g m^-2 year^-1 was higher than the one reported in this study^[9]. Conversely, the mean C_seq rates from the Solent’s seagrass meadows are similar to those reported by Jankowska et al., from Z. marina meadows in the Baltic Sea^[62]. The mean C_seq rates calculated by the conventional method, of 15.38 g m^-2 year^-1, were much lower than the mean CRS C_seq rate for the studied sites, and reported global ranges, suggesting possible underestimation. Moreover, an increase in C_seq rates over time is usually described by studies investigating restored seagrass sites^[13,67]. For example, Marba et al. reported an increase in C_seq rate from 16.2 ± 2.4 to 25.2 ± 4.7 g m^-2 year^-1[13], between 6-year and 18-year restored seagrass sites from Oyster Harbour, Australia, with C_seq rates values much closer to the ones calculated by the conventional method in this study. This is relevant since it has been reported that the seagrass meadows in this study are likely to still be in a recovery phase after reported dieback caused by wasting disease^[71,72]. Previous studies have highlighted historical limitations of global seagrass C_seq rate assessments, mainly based on limited data or the use of indirect approaches^[73,74]. Studies also call attention to the large, and yet understudied, variability in seagrass C_seq potential, and the need for preliminary and baseline regional assessments, such as the one presented by this study, despite their limitations^[74].

Influence of environmental variables and climate events on C_seq rates

The wide range in estimated C_seq rates by seagrasses is usually associated with their high net production variability, including variability among species and habitats, and methods used^[6,7,11,55]. However, some studies suggest that the variability in C_seq rates may be a consequence of difficulties in estimating reliable long-term accretion rates from these environments, with most reviews usually being dominated by species like Posidonia oceanica and Zostera marina^[3,55,75]. Moreover, environmental factors, extreme weather events, and anthropogenic impacts can influence C_seq rates between sampling sites in different ways. For example, an analysis of the CRS accretion rates per calculated date for the LGST 2 core in Hayling Island showed a large peak in accretion around 1970-1980, which did not correspond to a similar peak in C_seq rates. This suggests that there was a large input of mineral sediment during that period, possibly due to coastal erosion from a reported storm (1979), recorded coastal flood events (1981), and natural littoral drift in the area^[40-41].

Both cores from Creek Rythe showed very similar patterns of sediment accretion and C_seq rates over time, with peaks around the 1980s when calculated by the CRS method. These could be related to the reported periods of high wind speeds, as well as the occurrence of storms and floods during this period^[76]. However, cores from the other sites showed variations in this pattern; for example, one of the cores from Farlington Marshes (FMST 2) showed two peaks in C_seq rates around 1995 and 1955, even though sediment accretion rates were declining during these periods. These fluctuations could be related to the increase in %mud recorded for the same period, around 1995, preceded by a decrease around 1955, with closely related sorting patterns. This would also corroborate the notion that seagrass meadows in central southern England are still in a recovery phase and have not attained the levels of canopy density and carbon storage capacity recorded before the 1950s global dieback^[72]. Porchester’s PMST 1 core showed a steep increase in C_seq rates over the last decade, which is likely to be related to the higher % of mud found in the top layers of the core and an increase in allochthonous organic matter input by anthropogenic activities including urban development.

Several studies have reported evidence of climate-induced shifts in the seasonal growing pattern of marine primary producers^[32,77-80]. This would be increasingly probable if a cyclone or severe storm passed over already highly fragmented seagrass beds^[29,81,82]. Shields et al. related this shift to an increase in the length of the stressful summer season, impacting the timing of Z. marina’s growth and decline periods^[34]. Gentle increases in temperature enhance the rate of photosynthesis and primary production, which could promote seagrass growth, increasing meadow extent and, therefore, trapping more suspended particles, increasing C_seq rates^[83-86]. Conversely, temperature stress on seagrasses might result in distribution shifts, changes in patterns of sexual reproduction, altered seagrass growth rates, and metabolism, and changes in their carbon balance, which will cause plants to die^[1]. Lower temperatures, however, could negatively impact the seagrass growth rate, promoting dieback during wintertime, which could then increase C_seq rates by increasing the input of autochthonous carbon^[84,87,88]. An adverse impact of heavy rain, associated with storm events like the ones described in this study, is the decrease in salinity levels due to increased freshwater runoffs, causing large-scale losses of seagrass habitats that might take several years to recover^[89,90]. Anthropogenic or natural disturbance of seagrass meadows may decrease canopy height and density by damaging plants, as well as uprooting plants in some parts of the bed, resulting in the fragmentation of the meadow into patches^[91]. After an intense storm, it is typical to find dead seagrasses piled up in extensive wrack lines along the shore^[75]. Seagrass loss also results from smothering by sediments and light limitation due to increased turbidity from suspended sediments^[78]. These dieback events, promoted by extreme weather, can temporarily enhance sediment C_seq rates, due to the increase in organic matter from seagrass decay within the meadow and neighbouring areas^[23]. In contrast, if disturbances are too intense or prolonged, seagrass fragments may keep decreasing into smaller units until they disappear completely^[91]. However, due to their unexpected nature and unpredictability, the consequences of extreme climate events on seagrass communities can be incredibly challenging to monitor^[91], highlighting the need to monitor seagrass meadows regularly to measure their status and record short and long-term trends and possible effects of disturbance on these ecosystems, to establish reliable C_seq rates^[20,76].