This study gathered articles published between 1^st January 2000 – 1^st January 2020 to cover most ocean acidification (OA) research articles (Riebesell and Gattuso, 2015). Web of Science and Scopus were used to find articles covering climate change impacts on native European benthic marine organisms using the keywords: “ocean acidification”, “acidification”, “ocean warming”, “climate change”, “global warming”, and combinations/permutations of these. The results were snowballed following recommendations made by Orr et al. (2020); article lists from previous comprehensive reviews (Harvey et al., 2013; Kroeker et al., 2013) and their reference lists were used to locate additional studies. Article collection occurred between March 2019 – Jan 2020, article list available in Supplementary Data .

Data were limited to native European species ( Figure 1 ), excluding invasive species. Minimum standards for reporting of meta data were: control condition, acidification and or warming manipulated individually or synergistically within possible for conditions for 2100 conditions in European seas. Plausible conditions were determined based on author knowledge of likely conditions at a study site or species might experience, allowing inclusion of local contexts not identifiable in large, averaged projections. We exclude unrealistic stress (i.e., conditions beyond high end emissions scenarios by 2100) experiments in the context of climate change designed to improve our understanding of physiology. Studies assessing interaction with other stressors (such as heavy metal pollution or nutrients) were included if they analysed the climate stressors separately. Studies examining multiple species, biological responses, seasons, or geographical locations were included if they could be disaggregated.

Data on calcification, growth, metabolism, reproduction, photosynthesis, and survival were extracted ( Supplementary Data ). When the same biological trait was measured in multiple ways (e.g. growth as organism length and biomass), one was chosen at random following standard methodology (Harvey et al., 2013; Kroeker et al., 2013; Sampaio et al., 2021). To avoid higher weighting, one parameter was chosen when multiple biological responses (e.g. length and biomass to determine growth) were reported. Survival data were converted from mortality for standardisation. Many identified articles reported survival rates. However, these where often reported as percentages and thereby did not provide variance, preventing us from computing an effect size consistent with the rest of these meta-analyses, resulting in exclusion from our study. Time-series data were attributed to the endpoint of that experiment.

Datapoints, a form or variance (standard error, confidence intervals, or standard deviation), and sample size were extracted directly from either text or tables. Graphs were data mined using WebPlotDigitizer (Rohatgi,1 2019, See SI). Acclimation period (experimental period whereby organisms were transitioned to climate stressors e.g., increasing water temperature by 0.5°C daily to reach temperature increase value), study duration (length of organism exposure to climate stressor excluding acclimation period), organism developmental stage (categorised as Egg/larva, Juvenile, and Adult), pH change or temperature change, organisms’ size were extracted when possible. Size at the start of the experiment was converted to grams [g] or millimetres [mm]) based on the sample mean and log transformed. Insufficient data were extracted to allow for taxonomic-biological trait responses to organism size, so data were pooled combining biological traits to identify general trends on traits alone.

All analyses were conducted using the R (version 4.0.0; R Development Core Team). The package “Metafor” (Viechtbauer, 2010) was used to calculate effect size using the “escalc” function and conduct the meta-analyses. Log Response Ratio (LnRR) was chosen to measure effect size (Koricheva et al., 2013).

Multivariate meta-analysis models (function “rma.mv”) were used to calculate the mean effect and sampling variance of a treatment on a biological response. Random intercepts of article and species for each treatment were incorporated to account for possible results correlation due to studies from the same article or organism and account for pseudo-replication. Data were subsetted into taxonomic groupings, calcifiers vs non-calcifiers, heterotrophic vs autotrophic, and developmental stage when possible. Climate stressors were used as model moderators (mods = stressor -1). A result is considered significant if the 95% confidence interval for the overall mean calculated by a model does not overlap zero. Tests for residual heterogeneity (Q_E) were used to ascertain if additional moderators not considered in the analysis might be influencing study results (Hedges and Olkin, 2014).

To examine the impact of study duration and acclimation period we modified the multivariate meta-analysis models to include either of these as moderators within the models. E.g: [model <- rma.mv (yi, vi, method=“REML”, test=“t”, random = ~1|Article/species, data=data, mods = ~duration: stressor -1)].

Standardized diagnostic tests for multivariate meta-analytical methods are still evolving (Viechtbauer and Cheung, 2010; Habeck and Schultz, 2015) and currently only applicable for univariate models (e.g. models of class “rma”). To detect publication bias within our data we modified our multivariate models in accordance with Habeck and Schultz (2015) to calculate Egger’s regression. Egger’s regression detects funnel plot asymmetry, a common measure of publication bias, by looking if there are statistically absent studies based on the distribution of studies within meta-analysis datasets using linear regression (Egger et al., 1997). Multivariate meta-analysis models were extended to include the square root on effect size variance as a model moderator variable to perform a regression test (Habeck and Schultz, 2015). When this regression test intercept significantly deviates from zero, a data set is considered asymmetrical, thus biased (Sterne et al., 2005). Following previous recommendations (Egger et al., 1997; Habeck and Schultz, 2015) an analysis is considered biased if the intercept differs from zero at P = 0.10. Results from tests for residual heterogeneity (Q_E) and publication bias are reported in Supplementary Information .

Growth and reproduction of calcifying organisms in particular were reduced in response to warming and lowering of pH. However not all calcifying taxa experienced calcification reduction. Molluscs show negative impacts on calcification, growth and increases in metabolic costs, adding to the body of studies showing molluscs are a highly at-risk taxon globally (Parker et al., 2013; Kroeker et al., 2013; Goethel et al., 2017; Figuerola et al., 2021). Our findings contrast recent work on the Mediterranean Sea and a global analysis which suggested mollusc growth is unaffected by climate stressors (Zunino et al., 2017; Sampaio et al., 2021). Our results likely differs from the signal observed by Zunino et al. (2017) due to a larger sample size encompassing a wider range of taxa and growth responses. We interpret this apparent disagreement with the global analysis as indication that European mollusc species are particularly sensitive to climate stressors; a suggestion which is corroborated by global studies which have a strong bias towards temperate molluscan species in North Atlantic waters resulting in a similar interpretation (Kroeker et al., 2013; Harvey et al., 2013). European species have long been reported to be altering distributions in response to climate change (Helmuth et al., 2006) highlighting species at the edge of distributions are sensitive to climate stressors.

Reductions in growth rates will not only affect size and fitness, but also reduce molluscs ability to form biogenic structures, limiting their ability to support biodiversity (Strain et al., 2021). Many habitats created by these taxa are protected due to the associated high biodiversity (Donnarumma et al., 2018). Evidence suggests high food availability can buffer against climate stressors (Thomsen et al., 2013), but primary productivity globally is predicted to decrease by ~8.6% under business-as-usual scenarios (Bopp et al., 2013) thereby potentially amplifying the impacts depending on local habitat and other inputs into coastal ecosystems.

Reductions in mollusc growth rates and survival paired with reductions in crustacean survival and reproduction would impact commercial fishing in Europe by lowering catch potential and impacting livelihoods of European fishers. France, the Netherlands, and Spain annually produce approximately half of consumed European molluscs (Narita and Rehdanz, 2017) suggesting these regions are particularly susceptible to the economic repercussions of climate impacts on mollusc species. Additionally, the resilience of mollusc and crustacean populations against climate change has been reduced by destructive fishing and pollution for decades (De Groot, 1984). There are attempts to restore habitats in line with historical distribution, e.g., recreation of oyster beds around the UK (Harding et al., 2016) or red gorgonian in the Mediterranean (Linares et al., 2008). Restoring historical habitats that are reliant on calcifying organisms as keystone species may be a risky approach and not provide longevity due to observed impacts on growth rates, reproduction, and survival. Assessing local environmental conditions, the vulnerability of specific taxa, and options for reducing secondary local stressors are all needed to improve the long-term viability of these habitats.

Other ecosystem builders are similarly impacted, increasing the risk for coastal biodiversity. The decrease growth rates of calcifying algae are supported by other studies, highlighting the vulnerability of this group (Brodie et al., 2014; Chan et al., 2020). As they lose suitable environmental conditions (Simon-Nutbrown et al., 2020), calcifying algae are predicted to experience increased breakage due to increasingly intense weather, reducing structural complexity of Mearl beds (Melbourne et al., 2018). The reduced growth rates of calcifying algae suggested in our analysis will compound these problems (Ragazzola et al., 2013; Marchini et al., 2019). Furthermore, future sea level rise and changes in riverine input into coastal setting due to changes in precipitation and land use all threaten light attenuation in some habitats, for example where habitats cannot migrate landwards due to artificial coastal structures or cliffs, which exacerbates the impacts.

Our data suggest European corals are mostly unaffected by climate stressors, likely because most studies investigated the cold-water coral Lophelia pertusa which have been found in experiments to be resilient to projected end of century ocean conditions (Form and Riebesell, 2012; Wall et al., 2015; Büscher et al., 2017; Varnerin et al., 2020; Wang et al., 2021). While symbiont bearing tropical coral ecosystems are projected to be lost with only small amounts of additional warming (Hoegh-Guldberg et al., 2017), temperate deeper-water taxa appear more resilient (Kornder et al., 2018, this study). However, the 3D structure generated by deep water corals will be impacted by dissolution after death and thereby reduce the habitat and associated biodiversity (Hennige et al., 2020).

Our analysis suggests that autotrophic species, in our dataset mostly fleshy algae, will benefit from climate change. We echo the findings that algal species can display resilience to climate change stressors, in both regional (Strain et al., 2015; Phelps et al., 2017) and global analyses (Krumhansl et al., 2016). This finding is supported by field observations of intertidal habitats in Scotland which have observed increases in canopy-forming algal species (Burrows et al., 2017). However, the focus on the physiology masks species interactions which suggest that temperate algal communities may undergo simplification with decreases in structural complexity and diversity as climate change amplifies (Brodie et al., 2014; Agostini et al., 2021; King et al., 2021). Algal species frequently grow in high energy environments which experience rapid daily variations in tides and temperature. Therefore, they have evolved rapid growth responses to compensate frequent breakage in these environments (Rai and Gaur, 2012) which may also benefit under global change conditions. Due to their apparent resilience, associated ecosystem services like helping shoreline stabilisation and nutrient cycling may continue despite warming and acidification. However, the complexity of these systems is overlooked by our analysis of individual taxa as associated epifaunal species might be negatively impacted and thereby reduce biofim production on the algae (Berner et al., 2015). It is important to note though that eutrophication, pollution and habitat destruction from shoreline protection, fishing or future sea level rise were not part of this assessment. This apparent resilience may support endeavours to grow kelp or seagrass for carbon sequestration (Unsworth et al., 2018; Bach and Boyd, 2021), or grow algae as part of the blue economy (Araújo et al., 2021). Our results suggest efforts to restore algal habitats, such as kelp beds around Norway (Hynes et al., 2021), might benefit from the sugested resilience to climate stressors.

Our findings identify gaps in the experiments on some European benthic organisms. For example, while there is a high abundance of experiments on mollusc growth rates comparatively less is known for crustaceans or calcifying algae. This bias is replicated on a global scale (Kroeker et al., 2013; Harvey et al., 2013; Sampaio et al., 2021). Limited work on polychaetes and bryozoans hindered inclusion in our analysis expect for the generalised categories (calcifiers vs non-calcifiers, heterotrophs vs autotrophs). To gain a more differentiated understanding of what European seascapes may look like in the future and which services they can provide, greater efforts need to be taken to determine impacts on less studied taxonomic groups. Despite a clear importance of seasonal variation of environmental conditions, for example marine heat waves (Oliver et al., 2018), most studies were performed under stable environmental conditions resulting in insufficient data to analyse potential seasonal impacts on organism physiology. Such extreme conditions and their projected increasing frequency will threaten communities as the reducing time periods between these extreme events reduce opportunities for communities to recover (Bednar-Friedl et al., 2022).

The observed increases in algal metabolism in our analysis show an increased demand for oxygen, indicating that climate change is increasing standard physiological rates (Pörtner, 2008). Whether these increases are beneficial or detrimental depends on the relative deviation from optimal environmental conditions for temperature and oxygen (Pörtner, 2002; Pörtner and Farrell, 2008). The detected impacts on the metabolic responses of calcifiers corroborate Lefevre (2016) who found calcifying organisms consumed significantly more oxygen when exposed to climate stressors. In contrast, previous meta-analysis suggested that ocean acidification does not impact metabolic responses (Kroeker et al., 2013; Sampaio et al., 2021) while warming increased them (Sampaio et al., 2021). One potential interpretation of this difference is the nature of the regional marine environment. European waters experience large temperature fluctuations and organisms in these environments have shown to have limited ability to acclimate to climate stressors (Seebacher et al., 2015). Increased physiological demands can be compensated, for example by higher food intake at warmer temperature or to facilitate calcification at lower pH. However, future oxygen losses in the ocean (Schmidtko et al., 2017) could prevent organisms increasing their metabolic rates sufficiently. Therefore, interpretations on metabolic responses must be viewed in a wider biological context as increases in metabolism may result in negative responses in other parameters like growth rates (Kordas et al., 2011) and depend on the relative change in environmental conditions relative to critical and optimum values at their respective life stages (Pörtner and Farrell, 2008).

Organism size has been shown to greatly influence ectotherm survival (Peralta-Maraver and Rezende, 2021). Dynamic energy budget research shows organism metabolic rates generally decrease concurrently with size (Ren and Ross, 2001; Agüera et al., 2015). Increasing organism size decreases surface area to volume ratio, and oxygen uptake generally decreases relative to size as organism size increases (Lefevre et al., 2017). In general, metabolism increases in response to climate stressors. Meta-regression revealed that increasing organism size leads to negative metabolism responses ( Figure 3 ), but this is likely an artifact of established metabolic theory. This result highlights the complexity of impacts on organisms, and potentially explains discrepancies between results of different studies. We were unable to explore this in greater detail due to studies infrequently reporting organism size and encourage experimentalist to make this part of their data reporting.

Study duration (the period an organism was continuously exposured a to climate stressor) clearly impacts organism physiological responses to climate change. Using meta-regression to examine how study duration period changed trait responses to climate stressors suggests that impacts amplify the longer organisms are exposed. We expect effect size signals to shift towards zero if acclimation would occur. Our data though for growth and metabolism generally increased their response the longer experments were conducted ( Table 1 ). We suggest that European benthic organisms have limited capacity for acclimation over the time of the experiments. While we also detected many statistically insignificant signals, it is important to note that this does not indicate acclimation; it only suggests that study duration did not have a statistical relationship with effect size.

The mean experimental duration period across our data was only 86.3 days representing short-term exposure in contrast to climate change over decades. Long term experiments assessing acclimation are rare and typically conducted on organisms with fast reproduction time such as Pseudokirchneriella or Emiliania huxleyi (Schlüter et al., 2016; Limberger and Fussmann, 2021). Study length has not increased much since the meta-analysis Kroeker et al. (2013) clearly limiting our knowledge of the true long-term ecological impacts. Our selection of organisms may have a bias towards species which are adapted to deal with acute stress such as exposure to air, and highly variable environments. For example, photosynthetic performance of Fucus serratus in Spain when exposed to sudden thermal stress recovered within 24hrs, while populations from Norway, Denmark, and France did not recover (Jueterbock et al., 2014). Our results raise concerns that even organisms currently considered resilient and associated biological traits like algal growth rates can be overwhelmed by prolonged exposure to climate stressors. Chronic stress has been documented to lower the acclimation ability of marine organisms, reducing the upper thermal tolerances in mussels (Sorte et al., 2011) and coralline algae (Page et al., 2021).

Experiments used in this meta-analysis maintained stable conditions. Climate change will be associated with increasing frequency and duration of extreme events. Marine heat waves have led to mass mortality (Garrabou et al., 2009; Smale et al., 2019) and ecosystem restructuring (Wernberg et al., 2016; Ainsworth et al., 2020). Such extreme events have the potential to alter many currently statistically neutral responses) and are projected to increase in the future with climate change (Ranasinghe et al., 2021), thereby reducing time intervals which facilitate recovery (Cooley et al., 2022). Compounded by destructive fishing (Pedersen et al., 2017) and habitat fragmentation this increasing complexity of the currently observed risks (Simpson et al., 2021) highlights our still limited ability to project the true extent of the challenges marine ecosystems will face.