Our results revealed that herbivores were the most sensitive trophic level compared to predators and primary producers, which, in contrast, were highly tolerant to OA and OW stressors and their interactive effects. These results support previous findings of a meta-analysis carried out by 6, despite this study using only one-third of the number of species and studies compared to the previous analysis. The main reason for the smaller number of studies included here was because this meta-analysis only included fully factorial designs of multiple stressors. Although the combined effect of OA and OW did not have significant effects among trophic levels, the arrangement of the mean effect sizes was however once again often consistent with previous research about marine trophic levels and tolerance against climate change for individual stressors. For example Hu et al.6,31demonstrated a pattern, similar to the results presented here, whereby the effects of OA and OW were greatest on herbivores while higher trophic levels demonstrated greater tolerance. However, it is important to note that the combined effects of OA and OW were much lower than their individual effects (OA or OW), with the interaction effect often positioned around zero. This was a somewhat surprising result compared to what experts have predicted about multiple climate stressors and their potential to induce synergistic interactions (see discussion below).
Concerning primary producers, we received a somewhat unexpected result in which OA had a positive effect. However, this result may be explained by the fact that high CO2 levels in the surrounding environment can be utilized as a resource and have the potential to increase the carbon fixation rates in some photosynthetic primary producers 34–36. Yet, the combined effect of OA and OW was negative, where the low effect size from the elevated ocean temperature seems to act antagonistically, reversing the individual effects to an even greater negative effect. The reason for this result is at this point difficult to disclose, but a similar result was detected for both calcifying and non-calcifying primary producers (Fig. 3). The small effect from OW was also to some extent a surprising result, as related studies have demonstrated that primary producers thrive in high concentrations of CO2 and warm ocean temperatures. Still, this may involve only certain species of primary producers as 37 found that under the “future ocean” regime, larger chain-forming diatoms became dominant at the expense of smaller pennate forms. Such disparities across taxa may be what is detected in Fig. 2, where elevated ocean temperatures (OW) produce to some degree large variation around the effect size (see blue symbols). In conclusion, these results suggest that there is likely to be interspecific variation in the sensitivity of primary producers to climatic stressors.
Out results also support the hypothesis that calcifying herbivores, such as molluscs and echinoderms, are more sensitive to OA 6,10,38–40. Interestingly, for herbivores, the combined impact seemed less detrimental compared to the individual effects of OA and OW, and this is irrespective of calcifying or non-calcifying species. The compensatory effect of OW could possibly be explained by the warmer temperatures, which can increase calcium carbonate precipitation kinetics and offset the reduction in calcification caused by OA, a process observed in e.g. corals 41. An alternative explanation is that higher water temperatures benefit the development of species (Fig. S2), thereby reducing time in the vulnerable planktonic and early benthic juvenile stages that are particularly sensitive to stressors 42–44. In addition, studies have also suggested that ocean acidification and elevated temperatures impact different metabolic pathways, and in these cases, temperature was the overriding factor, particularly when focusing on mortality 45,46. Accordingly, herbivores seem more sensitive to OA and OW compared to primary producers when examining individual stressors in isolation. However, the interactive effect on herbivores was less severe, with effect sizes equal to primary producers, and lower compared to the predator groups (Fig. 2), where the main drivers were the combined effects on reproduction and survival (Fig. S2).
We found that predators was relatively tolerant to OA and OW 6,31 but also against the combined effect. This suggests that tolerance of predators to one stressor (OA/OW) may confer tolerance to another (OW/OA) stressor when both stressors act on the same physiological or ecological processes, or action pathways of stressors interact 47. For example, a study in a bony omnivorous fish demonstrated that exposure to an acute sublethal elevated temperature can lead to increased tolerance to acidification challenges 48 due to the linkage between CO2 levels and the expression of the heat shock proteins Hsp70 and Hsp90. The similar magnitude of effects by individual OA, OW and their combination suggests that the co-tolerance may be prevailing among higher trophic levels in response to multiple climatic stressors 49,50.
Our study represents the first meta-analysis to reveal trophic differences in response to multiple climatic stressors, where higher trophic levels seem to be more tolerant to climate stress than lower trophic levels, a pattern which partly has been confirmed for single environmental stressors by two previous studies6,31. However, other species-specific trait characteristics may additionally contribute to marine trophic differences in response to climate change, such as variation in body size 51–53, functional groups within trophic levels 30, differential ability to control body status and physiological processes 54–56, and different activation energy values and metabolic rates among trophic levels 21,57,58. However, none of these variables are likely to individually drive the trophic differences, instead they may interact together to contribute to the observed variation59.
Overall, our meta-analysis showed that synergistic interactions of OA and OW are much less common (16%) compared to antagonistic (44%) and additive (40%) interactions on marine species. Our results contradict previous results from a meta-analysis by 39, where the authors found that synergistic interactions between ocean acidification and ocean temperature dominated in marine ecosystems 60. The disparity between the studies could likely be due to difference in selection criteria of literature, where in this meta-analysis we only included fully factorial experiments. On the other hand, our findings are consistent with other recent reviews 24,61–64, indicating far fewer synergistic interactions than previously thought. It was also evident that the proportion of synergistic interactions decreased with increasing trophic rank, from primary producers (17%), herbivores (17%), meso-predators (10%), to no synergistic interactions detected among top-predators. The reduction in synergistic interactions, which principally are viewed as detrimental 62, while moving up the food-web may further support previous results that higher trophic levels may be less sensitive to climatic stressors than lower trophic levels.
There is a growing interest in how climate change impacts on marine organisms change along latitudinal gradients 3,65. It is, for example, widely known that biodiversity is more pronounced in the tropics and species richness declines with increasing latitude 65–67. The pattern is evident in both terrestrial and marine realms, and is strongly correlated with temperature 68,69. A study 65recently showed that species richness has declined around the equator, particularly in latitudinal bands with average annual sea surface temperatures exceeding 20°C 70. For elevated ocean temperatures, we confirm the findings of recent studies in that the relationship between absolute latitude and main effect size were all positive in the direction of higher latitudes (Fig. 4). Tropical species may be more sensitive to elevated temperatures since they already live close to the upper limits of their temperature tolerance 3. What was additionally interesting, was the apparent increase in variation of effect size found at 30 degrees N/S and at 45 degrees N/S, particularly for primary producers and herbivores (Fig. 4). However, since a meta-analysis is based on correlative statistics, we cannot resolve causality. It is nevertheless still intriguing to reflect on the fact that these latitudes represent environmental shifts from the tropics to the subtropics, and finally the temperate zone. As the tropics and subtropics are expanding poleward, different marine organisms will struggle more or less in order to respond to the stress of climatic change, either through their ability for rapid adaption, adaptive phenotypic plasticity or migration capacity71. With that said, it is important to remember that the mid-latitudes (between 30 degrees N/S and 60 degrees N/S) incorporate Australia in the southern hemisphere, and North America, as well as Europe in the northern hemisphere, and the simple fact is that most research grants and research projects are concentrated to these specific regions.
It is commonly advocated that conservation actions should focus on local stressors, as global stressors are often difficult to control and intervene 72,73. However, to implement effective mitigation strategies, it is key to identify stressor interactions to understand which stressors to act on, and when and where to intervene for prioritizing conservation actions 72,74,75. For that reason, we propose that research considering trophic levels will be an informative predictor and ecologically relevant for the future conservation actions, particularly, when species interactions are involved 59. Many research studies have promoted the need for multiple stressor assessments and where the combined effects of OA and OW will instigate greater negative effects than just the sum of the two (i.e., synergistic effects 22,62). In this meta-analysis, we examined 486 observations from 162 fully factorial experiments, which should be considered a large investigation with a particularly high statistical power. It is therefore notable to observe that the interaction effect between OA and OW often produced weaker negative effects compared to single-factor effects, and sometimes, reduced the negative effect, with additive and antagonistic interactions dominating. However, since it appears that trophic levels from other environmental realms 19,30,76 can respond differently to individual stressors, the results presented here may be exclusive to the marine realm and for that reason, multiple stressors deserve additional research and a deeper understanding.