Introduction

Increasing anthropogenic greenhouse gas (GHG) emissions are changing our planet’s climate and causing losses of biodiversity, ecosystem function, and ecosystem services worldwide1,2,3. Climate change is altering temperature regimes, and increasing the frequency, intensity and duration of atmospheric and marine heatwaves4,5,6,7. Intertidal habitats in estuaries and along the coast are particularly vulnerable because they submerge and emerge (being alternately exposed to seawater and air) during every tidal cycle8. Situated at the land-sea margin, they will also bear the impacts of increased storm events that deliver increased sediments, nutrients and pollution from land and larger waves and tidal forces from the sea.

Temperature modulates all physiological processes and thus governs the functioning and biogeochemistry of ecosystems, including the carbon cycle9. In general, metabolic rates of organisms increase with temperature, but this is constrained by thermodynamics and the thermal sensitivities of particular species10. Climate change is increasing mean temperatures, but extreme events and increased temperature variation are likely to drive ecosystem change as organisms reach their thermal tolerance limits11,12. Short-term extreme events such as heatwaves are strong drivers of ecosystem dynamics and resilience13,14,15 and are known to have long-term irreversible consequences for coastal populations, communities and their ecosystem functions [e.g. refs. 16,17]. Cumulative effects of prolonged heat exposures are expected to differ from effects of single short-term heat exposures18. Consequently, there have been many calls for studies that provide a deeper understanding of these impacts [e.g. ref. 4]. Intertidal organisms may be particularly vulnerable to extreme temperature events because they are already living close to their thermal maxima19.

Most research on GHG fluxes in intertidal coastal ecosystems has focussed on vegetated habitats (i.e., mangrove, saltmarsh, and seagrass for their ‘blue carbon’ potential20,21). In contrast, relatively little is known about the role of ‘unvegetated’ (microphytobenthos dominated) intertidal flats in the carbon cycle22, and how they will be influenced by higher frequency and intensity of extreme temperature events. Interest in the effects of changes in GHG emissions in intertidal habitats with broad areal coverage is driven by the potential for reinforcing feedbacks. If GHG effluxes (CO2, CH4) increase with increasing temperatures, this could create a feedback that contributes to accelerated climate warming23.

Intertidal flats may be a significant reservoir for carbon22, but their GHG emissions may offset their rates of carbon uptake24,25. The contributions of estuaries in global GHG flux calculations are often limited to water-air fluxes, i.e., excluding sediment-air fluxes in the intertidal zone26,27,28. For many estuaries, the contributions of intertidal sediments as carbon sinks remain uncertain, hampering efforts to constrain carbon budgets in estuaries with large proportions of intertidal area29. A further complication is that intertidal environments can vary greatly in sediment and habitat types present, tidal ranges, exposure times, and freshwater inputs that are likely to influence GHG fluxes.

Photosynthesis by microphythobenthos is a carbon sink (i.e., CO2 fixation). However, it also generates oxygen as a byproduct, which increases rates of organic matter remineralisation and CO2 release30 relative to other oxidants. Organic matter remineralisation continues until available oxidants are consumed or all oxidisable organic carbon is utilised. Methane is primarily produced (methanogenesis) in anoxic sediments and is the final step in organic matter degradation producing CH4 and CO230,31. Elevated temperature can be detrimental to microphytobenthic biomass and productivity, and can cause changes in community composition32. Photosynthetic capacity of microphytobenthos can increase with temperature33 but may be inhibited at temperatures greater than 25 °C34 or 35 °C33. Temperature influences CH4 production through direct effects on methanogens, and via indirect effects on carbon mineralisation and substrate availability (because mineralisation increases with temperature, which results in more carbon substrate available and therefore faster depletion of other electron acceptors)35.

There is a lack of data and understanding of the drivers of CH4 and CO2 fluxes in estuarine systems, especially spatial and temporal variability and dynamics36,37. GHG fluxes on intertidal flats are known to have diurnal patterns, and fluctuations in water level have a significant effect on CO2 and CH4 fluxes38,39,40. Fluxes of GHGs are influenced by temperature, generally because warming increases microbial activity38,40, therefore heatwave simulation is expected to alter CO2 and CH4 fluxes through direct and indirect effects on the microbial community, including the microphytobenthos. The influence of temperature on microbial communities is likely to have a large influence on the biogeochemical response to heating because thermal tolerance and adaptation will vary among species41, which will differentially impact different biogeochemical pathways. For example, net CH4 flux is the balance of activity of CH4-producing methanogens and CH4-consuming methanotrophs, but temperature may affect these processes differently30. Temperature increases have been shown to accelerate organic matter breakdown and increase GHG production42.

Changing sedimentary environments in estuaries, especially increasing bed sediment muddiness (silt + clay content), is the product of intensified anthropogenic land use, which can lead to changes in habitats and species composition, reduced ecosystem functioning, and loss of ecosystem services43,44,45. Consequently, sediment mud content is used as a proxy for estuary degradation46. Increasing sediment mud content can also reduce ecosystem interaction network complexity, which is an important indicator of ecosystem resilience47,48,49. Sediment characteristics, especially mud content, are known to be a key regulator of carbon fluxes in tidal flats50. Estuaries, at the interface of freshwater and marine ecosystems and subject to increasing sediment and organic inputs resulting from human activities, may be at risk of rapidly changing GHG emission status (switching from sinks to sources of CO251). Methane emissions increase as aquatic ecosystems become more impacted by humans (especially through increases in organic inputs), and they are generally higher in freshwater ecosystems than marine ecosystems31,52. In this study, we aim to evaluate the influence of low tide temperature extremes and subsequent impacts on GHG fluxes in unvegetated estuary intertidal flats with different sedimentary environments.

To investigate the influence of a low-tide atmospheric heatwave event on sediment-air fluxes of CO2 and CH4 we conducted a week-long field experiment using Open Top Chambers (OTCs) to increase low-tide sediment temperatures of two contrasting intertidal flats, one sandy and one muddy. We hypothesised that heatwave simulation would 1) increase the uptake of CO2 by stimulating photosynthetic activity of microphytobenthos (we expect surfaces to be warmed, and photosynthesis therefore enhanced, to a greater extent than within-sediment microbial respiration), and 2) increase the emission of CH4 by stimulating methanogenesis. We also hypothesised that 3) longer exposure to low tide heatwave simulation (i.e. cumulative days) would result in a greater effect on fluxes (i.e. effect size; the difference between control and treatment plot values). We chose heatwave treatment durations based on the definitions of atmospheric and marine heatwaves; events with at least three53 and five days54, respectively. The 5-day treatment duration was used to satisfy the temporal conditions of heatwave definitions, and the longer duration treatment (seven days) provided an extension of this, allowing us to begin to test for cumulative effects of slightly longer heatwave duration. We expected that 4) heatwave simulation would negatively affect the benthic macrofaunal communities through loss of sensitive species. Finally, we postulated that 5) treatment effects (and hypotheses 1–4) would be site-dependent where the muddy site would show less resilience to heatwave simulation than the sandy site.

Results

The OTCs used to experimentally alter temperatures within the sediment surface significantly increased both the mean and maximum temperatures during incubation periods (Supplementary Table 1, Figs. S1–S3). During incubations at the sandy site, sediment temperatures in control plots were on average 24.9 °C and increased maximally to 30.3 °C, and in treatment plots on average 26.3 °C with 36.1 °C maximum. During incubations at the muddy site, sediment temperatures in controls were on average 25.3 °C and increased maximally to 30.5 °C, and in treatment plots, sediment temperatures were on average 27.3 °C with maximums up to 36.3 °C. Mean sediment temperatures were greater at the muddy site (26.6 ± 0.15 °C) than the sandy site (25.8 ± 0.13 °C) but there was no difference in the maximum sediment temperature reached at each site. The strength of the treatment effect on mean and maximum temperature depended on Site, as indicated by a significant Site × Treatment interaction (Supplementary Table 1). The weather conditions varied throughout the experiment week, creating variability in the daily mean and maximum sediment temperatures measured in control and treatment plots (Figs. S1–S3, Supplementary Table 1). Similarly, the magnitude of the treatment effect on sediment temperatures varied throughout the week. Since the incubation days were offset between sites (i.e. Day 1 at the muddy site was a day later than the sandy site), the sandy site experienced six consecutive hot days followed by a warm day, whereas the muddy site experienced five consecutive hot days and two warm days (Figs. S1–S3).

Sediment properties were significantly different between sites, with higher sediment mud (Pseudo-F = 281, p = 0.0001) and organic matter content (Pseudo-F = 21.7, p = 0.0001) at the muddy site than at the sandy site (Fig. 1c, d, Supplementary Table 2). Seven days of daytime low tide heatwave simulation resulted in a reduction of sediment organic matter content at both sites (Pseudo-F = 27.5, p = 0.0001), but there was no evidence for this effect after only five consecutive days (Fig. 1c, Supplementary Table 2). Heatwave treatment also resulted in decreased biomass of Chl a (Pseudo-F = 7.12, p = 0.002), with greater effect magnitudes at the sandy site (Pseudo-F = 15.8, p = 0.0002, Supplementary Table 2, Fig. 1a).

Fig. 1: Sediment and macrofaunal community characteristics in control and heatwave treatment plots at each site.
Fig. 1: Sediment and macrofaunal community characteristics in control and heatwave treatment plots at each site.
Full size image

Sediment a chlorophyll a, b phaeophytin, c organic matter, and d mud content, and number of macrofauna e taxa and f individuals per core. Boxes display interquartile range with a black horizontal line showing the median. Whiskers show the non-outlier minimum and maximum.

Immediate and cumulative effects of heatwave on carbon flux

Sediments at the sandy site consistently showed uptake of CO2. Rates of CO2 uptake increased in the treatment plots over time (Fig. 2a–c, Supplementary Table 3), an effect that did not occur in controls plots (Supplementary Table 4). CO2 flux was not immediately (i.e. after two consecutive days of simulated heatwave) affected at the sandy site, but after five and seven consecutive days of simulated heatwave there was significantly more CO2 uptake by sediments in treatment plots relative to controls.

Fig. 2: Carbon dioxide flux in control and treatment plots at each site after 2, 5 and 7 days of heatwave treatment.
Fig. 2: Carbon dioxide flux in control and treatment plots at each site after 2, 5 and 7 days of heatwave treatment.
Full size image

Carbon dioxide flux in control and heatwave treatment (OTC) plots at the sandy site (a–c) and muddy site (d–f) after 2, 5 and 7 days of heatwave treatment. OTC boxes include data from both 5- and 7-day treatment plots for Days 2 and 5 (n = 20), and 7-day treatment plots only for Day 7 (n = 10). Note the difference in y-axis scales between a–c and d–f. Asterisks indicate significant differences between means at p <*0.05, **0.01, and ***0.001. Boxes show the interquartile range with a black horizontal line showing the median. Whiskers show the non-outlier minimum and maximum.

For the first 5 days of the experiment, sediments at the muddy site were emitting CO2 under natural conditions (Fig. 2d, e). The temperature treatment on average resulted in a switch to net CO2 uptake by the sediments as measured on Days 2 and 5. However, on Day 7 which was cooler than previous days (Figs. S1, S2), control plots were taking up CO2 and experimental warming resulted in higher rates of CO2 uptake. (Fig. 2f, Supplementary Table 3). Similar to the sandy site, control plot CO2 fluxes at the muddy site did not differ significantly by day (Supplementary Table 4).

The magnitude of difference (i.e. the treatment effect size) between control and treatment plot CO2 fluxes was significantly greater on Day 5 and 7 than it was on Day 2 at both sites (Fig. 3a, b). Repeated daytime low tide heatwave simulation resulted in a greater amount of CO2 uptake compared to controls at both sites, although the effect was less at the muddy site (Fig. 3a, b, Supplementary Table 3).

Fig. 3: Treatment effect magnitude for carbon dioxide and methane fluxes after 2, 5 and 7 days of heatwave treatment.
Fig. 3: Treatment effect magnitude for carbon dioxide and methane fluxes after 2, 5 and 7 days of heatwave treatment.
Full size image

Treatment effect magnitudes for carbon dioxide (a, b) and methane (c, d) fluxes at the sandy (a, c) and muddy site (b, d). Day 2 and Day 5 boxes include data from both 5- and 7-day treatment plots (n = 20), and Day 7 boxes include data from 7-day treatment plots only (n = 10). Boxes show the interquartile range with a black horizontal line showing the median. Whiskers show the non-outlier minimum and maximum. Asterisks indicate significant differences between means at p <*0.05, **0.01, and ***0.001.

Throughout the experiment at both sites, all plots were emitting CH4 (Fig. 4). Under ambient conditions (control plots) fluxes varied significantly by day and were greater at the muddy site compared with the sandy site (Supplementary Table 4). Heat treatment did not have a significant effect on CH4 fluxes at either site at any time during the experiment (Fig. 4, Supplementary Table 3).

Fig. 4: Methane flux in control and treatment plots at each site after 2, 5 and 7 days of heatwave treatment.
Fig. 4: Methane flux in control and treatment plots at each site after 2, 5 and 7 days of heatwave treatment.
Full size image

Methane flux in control and heatwave treatment (OTC) plots at the sandy site (a–c) and muddy site (d–f) after 2, 5 and 7 days of heatwave treatment. OTC boxes include data from both 5- and 7-day treatment plots for Days 2 and 5 (n = 20), and 7-day treatment plots only for Day 7 (n = 10). Note difference in y-axis scales between a–c and d–f. There were no statistically significant differences between treatment means on any day at either site. Boxes show the interquartile range with a black horizontal line showing the median. Whiskers show the non-outlier minimum and maximum.

Although there was no absolute effect of warming on CH4 fluxes, treatment effects varied with site and heatwave duration (Fig. 3c, d). At the sandy site, the treatment effect on CH4 emissions was negative throughout the experiment and did not change significantly with increasing days of heatwave simulation (Fig. 3c). At the muddy site, there were significant differences in the treatment effect size with increasing heatwave duration (Fig. 3d). Two days of low tide heatwave treatment resulted in greater CH4 emissions than control sediments, but after 5- and 7-days CH4 emissions were less than control plots (Fig. 3c, d).

Relationship between maximum sediment temperature and fluxes

For all plots combined, maximum sediment temperature did not affect CO2 flux, but site, date, and incubation day all had significant main effects (Supplementary Table 5, Fig. 5a). For CH4 flux, there were significant main effects of maximum sediment temperature, site, and incubation day (Fig. 5b, Supplementary Table 5). There was a significant interaction effect of maximum temperature and site on CO2 flux but not CH4 flux (Supplementary Table 5).

Fig. 5: Relationship between maximum sediment temperature and carbon dioxide and methane fluxes.
Fig. 5: Relationship between maximum sediment temperature and carbon dioxide and methane fluxes.
Full size image

a Carbon dioxide and b methane flux relative to maximum sediment temperature for all treatment plots throughout the experiment at the sandy (blue) and muddy (red) sites. Shading indicates 95% confidence interval.

Influence of cumulative heating on macrofaunal community structure

The macrofaunal community structure was distinct between the sites, the sandy site had a greater number of taxa per core than the muddy site and a similar number of individuals (Supplementary Table 6, Fig. 1e, f, Fig. 6). The sandy site was characterised by moderate densities of cockles (Austrovenus stutchburyi, ~7 individuals core−1) with other dominant taxa including Halopyrgus pupoides gastropods, and Ceratonereis sp polychaetes. The muddy site was characterised by the presence of burrowing crabs (Hemiplax hirtipes, ~1–2 individuals core−1) and other dominant taxa included Paracorophium excavatum (amphipod), and Arthritica sp.5 (bivalve). Experimental warming significantly altered the community composition with the effect being dependent on site, and a greater effect on the community composition at the muddy site compared with the sandy site (Fig. 1e, f, Fig. 6, Supplementary Table 6). At the sandy site, community composition on Day 7 was more similar to controls than that on Day 5. The direction of change in community structure for the 5- and 7-day treatments was similar at the muddy site, however at the sandy site, there was obvious divergence between the two treatment durations (Fig. 6). The simulated low tide heatwave did not affect the number of species, but the number of individuals present reduced in both the 5-day and 7-day duration treatments (Supplementary Table 6). Key bioturbating species were negatively impacted by the simulated heatwave; Paracorophium excavatum accounted for the greatest amount of dissimilarity between control and heatwave treatments at both the sandy and muddy sites (Supplementary Tables 7, 8).

Fig. 6: Macrofaunal community structure grouped by site and treatment.
Fig. 6: Macrofaunal community structure grouped by site and treatment.
Full size image

Bootstrapped metric multi-dimensional scaling plot of macrofaunal community structure at the muddy and sandy site in control plots (blue), and heatwave treatment plots after 5 (pink), and 7 (dark red) days of simulated heatwave.

Discussion

Our experiment manipulated in situ estuary low tide sediment temperature simulating a multi-day heatwave. A heatwave can be broadly described as “a period of consecutive days where conditions are excessively hotter than normal”, however there are many different atmospheric heatwave definitions and indices used which are usually specific to the region and impact of interest (e.g. human health, agriculture)53. Alongside metrics of frequency, intensity and extent, persistence is a key component that is used to define heatwaves; with increasing duration associated with increasing impacts6. We demonstrated that low tide extreme temperature conditions significantly alter macrofaunal community composition, sediment characteristics, and in turn CO2 and CH4 fluxes. We showed that the degradation status of estuary tidal flats (i.e. using sediment muddiness as a proxy46,47,48,49) influences their CO2 and CH4 source/sink status, and response to high temperature. Under natural temperature conditions, the sandy site was a greater CO2 sink and a smaller CH4 source compared with the muddy site. Five or seven days in a row of low tide simulated heatwave resulted in greater effects on GHG fluxes than just two days, indicating the potential for cumulative effects of climate-related warming on the functioning of intertidal estuarine ecosystems.

The simulated heatwave led to increased net uptake of CO2 by the sediments at both sites, possibly due to upregulation of photosynthesis, reduced respiration (by microbes and/or macrofauna), or a combination of both. In coastal wetlands, biofilms can reduce net CH4 and CO2 emissions42 and studies have shown intertidal flats to be net CO2 sinks with uptake related to Chl a concentrations50,55. Positive relationships between microphytobenthic productivity and temperature have been shown elsewhere with maximum values occurring around 30 °C56. However, in our study, microphytobenthic biomass declined in response to experimental heating inside OTCs, while photosynthesis appeared to have been positively influenced by temperature increases. The effects of temperature on microphytobenthos are species-specific, with photoinhibition and recovery of photoinhibition having different, species-specific temperature thresholds57.

Microbial growth and activity responses to temperature increases reported in the literature vary widely and mechanisms behind the changes are poorly understood58. Methanogenesis may be more temperature dependent than photosynthesis and respiration59. Similarly, methanotroph activity may be outpaced by that of methanogens under warming scenarios, and this effect may be greater in muddy sediments with a thin oxic layer60 resulting in increased CH4 emissions. Our study did not detect a significant treatment effect of warming on CH4 emissions, but analysis of maximum sediment temperatures across all plots showed a significant positive relationship with CH4 efflux. In freshwater ecosystems positive relationships between temperature and methanogenesis have been evidenced, however, CH4 oxidising organisms are not temperature sensitive to the same degree, resulting in changes in the balance between these processes that results in higher CH4 emissions60,61. The positive correlation between maximum sediment temperature and CH4 emission in our study suggests that with increasing frequency of warm summer events and increasing temperature maxima in intertidal estuary ecosystems, CH4 emissions may begin to trend upwards59,62. Methane oxidation occurs in oxic sediments, and methanogenesis in anoxic sediments, therefore, different sedimentary environments will have a different balance in these processes (i.e., muddy sediments may have a very thin oxic layer, whereas sandy sediments will have a deeper oxic layer). The observed differences in treatment effect sizes on CH4 and CO2 flux between sites, as well as the different flux–temperature maxima relationships between sites, suggests that consequences of heatwave events will be context dependent and potentially worse in degraded estuaries.

Simulated low-tide heatwave conditions changed the makeup of the macrofaunal community, an effect that was greater at the muddy site, and the observed cumulative effects on GHG fluxes may have been associated with these faunal changes. The significant declines in macrofaunal abundance in treatment plots could be indicative of mortality or behavioural responses that result in emigration from the treatment plots, but neither of these were observed directly. OTCs formed a seal at the sediment surface preventing surface emigration during incubation times, however subsurface emigration (i.e. burrowing and lateral movement) may have been possible for some mobile species although this was not observed. Faunal movement from treatment plots may also have occurred outside of the OTC incubation periods. While we cannot discern whether the decline of macrofaunal abundance was from mortality or emigration (or both), both suggest that the simulated conditions were unsuitable for those species, and that real-world heatwaves would have similar community effects. Future studies of macrofaunal functional trait responses to heatwaves will provide insights into the vulnerabilities and implications of future heatwave events for animals with, for example, different living habit (e.g. attached, burrower, free living, etc.) or mobility (e.g. limited, mobile, sessile) traits63. Temporal dynamics of heatwaves are important, with evidence that short periods of heating can enhance bioturbation while longer duration heating can cause mortality of bioturbators64. Macro-infauna, especially bioturbators, are known to increase production and emission rates of CO2 and CH465,66. There were contrasting patterns in the community response to the two heatwave treatment durations. At the muddy site the community structure shifted in the same direction indicating cumulative effects of increasing heatwave duration. In contrast, at the sandy site there was divergence between the community composition of the 5- and 7-day treatments, and the 7-day heat treatment plots were more similar to the control plots than the 5-day treatment. This could be due to the cooler days experienced at the end of the experiment (Days 6 and 7) which may have allowed some recovery and may be reflected in the CH4 flux effect size difference between Days 5 and 7 at the sandy site. At both sites there were reductions in the abundance of bioturbating amphipods, Paracorophium excavatum, in the treatment plots.

Increasing temperature was expected to increase macrofaunal respiration and increase CO2 emission rates55. For most organisms, metabolism increases with temperature up to a maximum threshold10. Reduction in macrofaunal respiration could partly explain the net increase in CO2 uptake if their metabolism reached this temperature maxima threshold. A parallel study showed evidence for metabolic stress of a key bivalve species, Austrovenus stutchburyi, following the temperature treatment67. Bivalve metabolism typically relies on oxygen generating a steady supply of energy to support various physiological processes68. Elevated temperatures can reduce concentrations of dissolved oxygen in sedimentary pore water and accelerate metabolic rates in bivalves, which leads to increased energy demands and may cause a shift from aerobic to anaerobic respiration pathways67. Indeed, the metabolic responses (i.e. metabolites and metabolic pathways) of this key bivalve to warming during this experiment were related to effects on energy metabolism, accumulation of stress-related metabolites, disruption of amino acid metabolism and impacts on lipid metabolism67. Loss of bioturbators following persistent high temperatures could also reduce porewater flows and the movement of substrates throughout the sediment profile (that fuel organic matter breakdown and methanogenesis) as well as diffusive effluxes of gas from sediment to the atmosphere42. The simulated heatwave may have caused sublethal effects on macrofauna such as reduced bioturbation rates, increased migratory behaviour or deeper burrowing64,69, which would in turn alter redox gradients, porewater flows, and diffusion rates of gases from the sediment.

It is important to consider our results in the context of the entire tidal cycle since physical and biological processes vary greatly throughout, and gas fluxes were only measured following OTC incubations during the daytime low tide. Fluxes of CH4 and CO2 from the sediment to the air (or water) occur through diffusion or ebullition30, processes that will be influenced by tide state and faunal activity. Changes in sediment biogeochemistry may have occurred that did not manifest as changes in gas fluxes during the daytime low tide when measurements were made. For example, incoming tides would have flushed out gases from pore spaces as would hydrodynamic forces that operate during high tide70. Furthermore, many intertidal species increase their activities during high tide which will enhance sediment-water coupling. Nighttime low tide air and sediment temperatures were not elevated in our study but it is possible that they would be in a real heatwave event, which in turn could exacerbate physiological and biogeochemical stress and therefore compound the effects seen in our study. Coupling of marine and atmospheric heatwaves is likely [e.g.71], and may reduce the metabolic reprieve provided by high tides72. Coincidentally, during our experiment there was a moderate marine heatwave on the east coast of New Zealand73 which may also have contributed to ecosystem stress and the cumulative effects we measured.

The increasing treatment effect size on CO2 and CH4 fluxes with increasing duration of simulated heatwave indicates that prolonged periods of temperature extremes will have consequences for the greenhouse gas budgets of estuaries. The cumulative effect of repeated daytime low tide heatwave exposure on gas fluxes was greater than the effects of a short-term (i.e. 2 days) atmospheric heatwave. Duration of heatwave events can determine ecosystem GHG emission response23. Effects of temperature treatment on CO2 fluxes increased with cumulative days with simulated daytime low tide heatwave and effects persisted on Day 7 even with the advent of cooler weather at the end of the experiment. The greatest influence of persistent warming relative to control values was for CH4 emission at the muddy site; two days of low tide heatwave conditions resulted in increased emissions, and seven days resulted in decreased emissions. This indicated a rapid upregulation of activity followed by acclimatisation of the microbial community that occurred over several days. Heat stress can cause changes in microbial community composition and functioning74,75. In particular, warming may influence the makeup of the methanogen community which can alter the pathways of methanogenesis60. As temperatures increase, functional redundancy of the microbial population declines75. This may explain the change in the magnitude of difference of CH4 fluxes between controls and treatments (i.e. effect size) observed over time at the muddy site: after 2 days of simulated low tide heatwave, the treatment effect on CH4 emissions was positive, whereas after five or seven days, the treatment effect was negative.

This research demonstrates differing effects of atmospheric heatwaves (and increasing temperature maxima) in different sedimentary environments which has implications for estuaries experiencing increasing sediment mud content. Muddy sediments tend to have higher organic content, shallow sediment oxic layers, and greater water content during low tide compared to sandy sediments, conditions that favour CO2 and CH4 production76. Other research has similarly shown that increases in temperature have greater effects in organic rich sediments and that this might help the recovery of over-enriched sediments by speeding up mineralisation of excess organic matter77. Along with sedimentation, eutrophication threatens coastal ecosystems, and muddy sediments tend to be more nutrient enriched, conditions that can further enhance CH4 emissions52,78.

We show that atmospheric heatwave events will manifest differently depending on the sedimentary environment, and that effects on benthic macrofaunal communities will be greater in muddy areas. This has implications for ecosystem resilience, since degraded ecosystems typically have lower species and trait diversity, and reduced interaction network complexity, making them more vulnerable to disturbance events such as heatwaves79,80,81,82. Benthic macrofauna are integral components of sediment biogeochemistry, and small changes in species composition such as reduced abundance of bioturbating organisms will have feed backs that negatively impact degradation-alleviating ecosystem functions such as denitrification83,84. To understand the consequences of climate change on biogeochemical cycling, there needs to be quantification of microbial processes and their environmental drivers58. There has been a call to action to include microbial measures and responses into climate change research, particularly regarding carbon and nitrogen fluxes58. Future studies on GHG flux response to extreme temperature events in coastal ecosystems would be enhanced by incorporating quantification of microbial communities with tools such as microbial multi-omics.

Climate change associated heatwave events coupled with degradation of coastal ecosystems will have substantial effects on macrobenthic biodiversity, biogeochemistry and GHG balance. Unvegetated intertidal flats were CO2 sinks in our study, and this was enhanced by warming. This effect was greater at the sandy biodiverse site compared to the muddy site. Degraded ecosystems subject to sedimentation tend to be a greater source of CH4 and CO2, and heatwave effect sizes will be greater and more variable than at more resilient sandy ecosystems. Climate change-associated environmental change will have a potentially major influence on macrofaunal communities and habitats, with serious implications for ecosystem functioning and GHG budgets in a future of increasing frequency, intensity and duration of heatwave events.

Methods

The low tide heatwave experiment was carried out in Waihī Estuary, Bay of Plenty, in Northern New Zealand from 16 to 23 February 2024 (Fig. 7). Waihī Estuary is a barrier enclosed estuary with a large proportion of intertidal area (57%), a catchment composed primarily of low-lying agricultural land use85. It has a high level of freshwater input through channelised agricultural drains, delivering high nutrient loads (488 kg nitrogen and 46 kg phosphorus per year on average) and fine sediment inputs (13.8 tonnes per year)86. The circulation in some parts of the estuary is restricted and some areas in the upper reaches can be described as hypoxic and degraded with high incidence of nuisance macroalgae. Deposition of terrestrially derived fine sediments (‘mud’) and poor flushing in the upper parts of the estuary has resulted in a gradient of sediment type (muddy to sandy).

Fig. 7: Map of the experimental site in Waihī estuary in New Zealand.
Fig. 7: Map of the experimental site in Waihī estuary in New Zealand.
Full size image

a Location of Waihī estuary (red box) in the North Island of New Zealand, b locations of muddy (red) and sandy (blue) sites within Waihī Estuary, c experiment configuration showing control (light blue), 5 day (pink) and 7 day (red) treatments plots at both sites, d birds-eye view of experiment configuration with Open Top Chambers at the muddy site, and e an experimental plot with Open Top Chamber and stake with temperature logger attached.

Experimental design

Two experimental sites with contrasting sediment types (sandy, muddy) ~200 m apart were established in the mid-reaches of Waihī Estuary (Fig. 7a–c). This area contained expanses of relatively sandy and relatively muddy sediment but did not exhibit any other signs of eutrophication. The duration of low tide emergence was approximately 5 h. At each site, a total of 30 plots were established with 10 replicates each for control, short duration (5 days of heatwave simulation), and long duration (7 days of daytime low tide heatwave simulation) treatments. Wooden stakes with temperature loggers (EnvLoggers, Electric Blue CRL) positioned 1 cm below the sediment surface were installed in each plot 2 weeks prior to commencement of the experiment. Each logger recorded the temperature every minute for the entire duration of the experiment. Replicates of each treatment type were arranged in a randomised block design in a 3 column × 10 row array (OTCs 5 m apart) at each site to prevent imbalances in sample allocation with respect to measured or unmeasured environmental heterogeneity across the sites.

Low tide heatwave simulation treatment

To elevate sediment temperature during tidal emersion, Open Top Chambers (OTCs) were placed on treatment plots for ~4 h during daytime low tides every day for five or seven consecutive days (short and long duration treatments, respectively) at each site. OTCs were conical (base dia. 0.8 m, top dia. 0.3 m), constructed of 1.5 mm thick polycarbonate sheeting and were deployed with the bottom edge flush with the sediment. OTCs provided passive warming of the treatment plot sediments but prevented convective cooling by the wind. Control and treatment plot solar radiation levels were similar (Fig. S4), and the conical open top design minimised increases in relative humidity. Temperature differences between control and treatment plots (i.e., the warming effect of the OTCs) were dependent on daily weather conditions. The greatest amount of warming (relative to controls) occurred on sunny days with a breeze, and the least occurred on cloudy, still days. Following low-tide incubations, OTCs were removed so there was no interference with high-tide (or nighttime low-tide) hydrodynamics, faunal behaviour, immigration or emigration of organisms. To enable time to conduct the sampling required at each site, the establishment of warming treatments was offset by a day between sites; sandy site OTC incubations were from 16/2/24 to 22/2/24 and muddy site OTC incubations were from 17/2/24 to 23/2/24.

Gas flux

CO2 and CH4 gas fluxes were measured in control plots every day at both sites and in OTC treatment plots on days 2, 5, and 7 at each site. Gas fluxes in control plots were measured first so that treatment plots could be measured following at least three hours of OTC incubation. Daily sampling of controls at both sites enabled us to account for weather and avoid confounding site and date in subsequent statistical analyses.

A Li-7810 portable trace gas analyser (LiCor, Nebraska, USA) and static chamber with opaque lid were used to measure fluxes of CO2 and CH4 between the sediment and the atmosphere (i.e. across the sediment-air interface). A circular stainless-steel chamber (40 cm dia., 22.9 L vol.) with a removable lid enclosed a 0.11 m2 area of sediment (Fig. S5). To reduce the influence of sediment disturbance on flux measurements, chamber bases were placed in the plots ~10 min prior to incubations by gently pushing them to a depth of 2 cm. We used a line drawn 2 cm above the lower edge of the chamber base to ensure constant depth and chamber volume between plots. We alternated three chamber bases to reduce the time between flux measurements at each plot to <1 min. Bases contained a trough filled with water to create a seal with the lid, which was connected via tubing to the gas analyser. The concentration of CO2 and CH4 was recorded every second for 4 min (measurement rate 1 Hz, flow rate 300 ml min−1). The Li-7810 gas analyser uses Optical Feedback-Cavity Enhanced Absorption Spectroscopy (OF-CEAS), which utilises the response in the infra-red of the gas being measured. Precision for CH4 at 1 Hz is 0.1 nmol mol−1 while for CO2 it is 0.05 μmol mol−1. The instrument was calibrated before the field campaign using four tanks with CH4 and CO2 amount fractions determined by the World Meteorological Organisation Central Calibration Laboratory; the CH4 is reported on the WMO X2004A scale and the CO2 on the WMO X2007 scale. The accuracy, determined through repeated determinations of a known gas for three-minute averaging, is 0.1 nmol mol−1 for CH4 and 0.1 μmol mol−1 for CO2.

Flux values were calculated using the goFlux package87 in R software v4.3.288. Flux peaks were manually identified using incubation start times and a 30 s deadband period. The goFlux function was used to calculate both a linear (LM) and non-linear (HM) flux estimate, and the best.flux function was used to automatically select the best flux estimate based on criteria scores for model fit (AICc, MAE, g.factor, MDF). In most cases, the HM model provided the best flux estimate. CO2 fluxes were expressed in mmol m−2 h−1 and CH4 fluxes in nmol m−2 h−1.

Sediment and macrofaunal sampling

At the end of the experimental treatment period (Day 5 for 5-day plots, and Day 7 for control and 7-day plots) cores were collected for analysis of sediment grain size, organic matter content, and microphytobenthic pigments (core size: dia. 2.6 cm, depth 2 cm) and benthic macrofaunal community composition (core size: 13 cm dia. 15 cm depth). Macrofaunal cores were sieved over 500 µm mesh, preserved in 70% isopropyl alcohol and stained with Rose Bengal. All individuals were counted and identified to the lowest possible taxonomic resolution (usually species). Sediment cores were frozen and stored at −20 °C. Wet sieving (after digestion in 6% hydrogen peroxide for 48 h) was used to measure the cumulative percentage mass of sediment size fractions89. Sediment organic matter content was quantified as loss on ignition (400 °C for 5.5 h) after drying to a constant weight (60 °C)90. The microphytobenthic biomass of the sediments (i.e., chlorophyll a content) was quantified by extracting chlorophyll a from freeze-dried sediments by boiling in 95% ethanol, then analysing the extract on a spectrophotometer (UV1800; Shimadzu, Kyoto, Japan)91. All sediment and biological samples were collected under the Fisheries New Zealand Special Permit 842-4 to Earth Sciences New Zealand (formerly the National Institute of Water and Atmospheric Research).

Data analysis

Sediment temperature data (from 1 cm below the sediment surface) were trimmed, such that we only analysed data from periods when the OTCs were positioned atop treatment plots at each site. Tests for differences in mean and maximum temperature between sites and treatments was carried out using Permutational Analysis of Variance (PERMANOVA) on normalised data, with Euclidean distance matrices and 9999 permutations92. Pair-wise post-hoc tests (9999 permutations) were conducted where main tests showed significant differences. Tests for differences in sediment properties (Chla, phaeophytin, mud content, organic content) were similarly performed.

To analyse the effect of heatwave treatment on fluxes, treatment plot flux data were combined (i.e., 5-day and 7-day plots) and separate PERMANOVAs were conducted for CH4 and CO2 fluxes for each site for Days 2, 5 and 7. Treatment plot data were combined because at day two and five all OTC plots were not statistically different. To understand the treatment effect on fluxes, the magnitude of difference between control (average for each site (i) on each day (k)) and heat treatment plots (i.e. effect size) was calculated using Eq. (1). To test if the immediate effect of warming was different to the cumulative effect of multiple days of heating, separate PERMANOVAs for each site were conducted comparing differences in flux magnitude between Days 2, 5 and 7.

$${{\rm{Flux}}} \; {{{\rm{magnitude}}} \; {{\rm{difference}}}}_{{{{\rm{Day}}}\,}_{i},\,{{{\rm{Site}}}\,}_{k}} \\ ={{{\rm{Treatment}}} \; {{\rm{flux}}}}_{{{{\rm{Day}}}\,}_{i},\,{{{\rm{Site}}}}_{k}}-{{{\rm{Average}}} \; {{\rm{control}}} \; {{\rm{flux}}}}_{{{{\rm{Day}}}}_{i},\,{{{\rm{Site}}}}_{k}}$$
(1)

To understand the role of sediment temperature on CO2 and CH4 fluxes, PERMANOVAs were conducted with maximum sediment temperature as a continuous covariable, and site, date, and OTC incubation Day as fixed factors. Maximum sediment temperature was used as we expected this to be a stronger driver of change than mean sediment temperature. Including the date allowed us to account for any differences between days that may have influenced the analysis (i.e., since the OTC incubations and measurements commenced one day later at Site 2 than Site 1). This analysis also allowed investigation of interaction between temperature and site.

To understand how experimental warming influenced the relationship between temperature and gas fluxes, we similarly performed PERMANOVAs with maximum sediment temperature as a continuous covariable with control and treatment data from both sites.

To explore the influence of enhanced warming on the macrofaunal community structure, a Bray-Curtis similarity matrix was created with square-root transformed data. We then used a bootstrapped metric Multi-Dimensional Scaling (MDS) ordination (50 bootstraps per group, minimum rho of 0.99) and a two-way PERMANOVA (9999 permutations) with Site (1, 2) and Treatment (Control, 5 day, 7 day) as fixed factors. Post-hoc pairwise tests were performed where significant interactions occurred (9999 permutations). A Similarity Percentage Breakdown (SIMPER) analysis was performed to analyse the species driving differences between treatments (i.e., which species increased or decreased the most due to heatwave simulation). All PERMANOVAs were performed as first described unless otherwise noted. PERMANOVAs, MDS and SIMPER analyses were carried out using Primer v7 software with the PERMANOVA add on93. All other data visualisations were conducted using ‘ggplot2’ from the ‘ggpubr’ package94 with R software version 4.5.188.