1.3.2.8 Marine ecosystems & environment

Climate change, pollution and overexploitation are affecting the marine environment at the physical, chemical and biological levels (e.g. Heinze et al., 2021; Jouffray et al., 2020; Bindoff et al., 2019). Pelagic marine ecosystems (defined as the water column from the surface ocean to the seafloor) as well as benthic marine ecosystems (defined as restricted on the seafloor) from the organism to the community level are changing at the same time as the ocean waters are becoming more warm, acidic and deoxygenated. In this section, we outline five potential tipping systems ranging from fisheries collapse and regime shifts in marine communities to ocean water hypoxia and the nonlinear weakening of parts of the ocean’s biological pump (Figure 1.3.24).

Evidence for tipping dynamics

Fisheries collapse

Over the past decades many fisheries have collapsed primarily due to over-exploitation, but they are increasingly threatened by climate change. Fish stocks are defined as management units of a species; thus one fish species can have multiple stocks (e.g. more than 20 stocks in the North Atlantic are assessed for Atlantic cod, Gadus morhua).

Among more than 200 exploited fish stocks, 23 per cent of the species showed at least one stock collapse (biomass below sustainable reference points) (Pinksy et al., 2011). Concerningly, 40 per cent of the collapsed stocks present different regimes of productivity (different relationships between fishing and biomass at different productivity stages) (Vert-pre et al,.2013) that potentially indicate the presence of regime shifts and hysteresis. But, while for some species there is clear evidence of regime shifts (Atlantic cod stocks), for others more studies are needed (Frank et al., 2016; Sguotti et al., 2019).  

Fish stock collapses can be due to different feedback mechanisms. The collapse of a stock can induce food web changes (i.e. trophic cascades) that, by modifying the other species of the community and their interactions, can maintain the population at a low level through predation or competition. For instance, large predators such as Atlantic cod may be successful because of the ‘cultivation effect’: adult cod prey on the juveniles of forage fishes (small pelagic fish which are preyed on by larger predators) that are competitors or predators of juvenile cod. Once the collapse in the biomass of cod occurs, the predation on the forage fish is released and these species start to thrive. Forage fish then prey on juvenile or recruit cod, thus maintaining the population in a depleted state. Examples of this particular dynamic can be found in Newfoundland and also the Baltic Sea (Walters and Kitchell, 2001). Another possible mechanism of hysteresis is the so-called Allee effect, which takes place when recruitment of a population (the process by which new organisms are added to a population) is positively correlated with its biomass. This means that a minimum population size is needed for the population to grow; otherwise it collapses. Thus, if biomass collapses, recruitment will also drastically decline, limiting the capacity of the population to recover. The Allee effect has been shown to be one of the possible hysteresis mechanisms of 13 stocks of Atlantic cod (Winter et al., 2023).

It is difficult to detect specific thresholds in fisheries in general, since every species and every stock within each species is impacted by different levels of the same driver and may experience different pressures. However, it has been shown that, for Atlantic cod stocks, the threshold was created by the combination of multiple drivers, especially warming and fishing (Sguotti et al., 2019; Beaugrand et al., 2022). Specific thresholds need to be detected for every stock. 

Beaugrand et al., (2022) have shown that rebuilding cod stocks may depend upon the fishing-environment interaction.

When the environment becomes unsuitable at the same time fish stocks collapse, rebuilding the stock may take time or even be impossible so long as adverse environmental conditions persist. This provides an explanation as to why, despite the fishing moratorium near Newfoundland, a partial recovery took more than two decades.

DFO 2018

Long-living, slow-growing species might be more prone to irreversibility. For instance, 16 out of 19 Atlantic cod stocks present regime shift dynamics due to fishing and warming and their recovery is hindered by the presence of hysteresis (Sguotti et al., 2019; Möllmann et al., 2022; Frank et al., 2016).  

Marine community shifts

Marine community shifts take place when abrupt changes cascade through several species or functional groups of an ecosystem, i.e. the change is not limited to a single species, as in a fish stock collapse, but can cascade all the way from top predators to phytoplankton (Figure 1.3.26).

Many community shifts have been reported in marine ecosystems (Conversi et al., 2015; Beaugrand et al., 2019; Möllmann et al., 2021; Ban et al., 2022; Sguotti et al., 2022). Some ecosystems have even experienced several marine community shifts, such as the Black Sea and Baltic Sea. In the Black Sea, the first major shift started in the end of 1960s with the overfishing of pelagic top predators, enabling surplus phytoplankton and jellyfish production during the following decades, and resulting in increased hypoxia (lack of oxygen necessary for life) followed by collapse of small pelagic fish and domination of jellyfish (Daskalov et al., 2017). In the Baltic Sea, the increased inflow of nutrients and organic matter resulted in the eutrophication of the main basins around the 1950s, enabling higher biological production, but also worsening hypoxia (Österblom et al., 2007).

Community shifts related to tipping responses mostly occur when the system is controlled by a few key species through trophic cascade (Beaugrand et al., 2015; Daskalov et al., 2007, 2017). Trophic cascades can be environmentally induced or induced by anthropogenic pressures such as pollution or overfishing (Casini et al., 2009). The mechanisms at the origin of the apparent synchronicities among marine community shifts have been debated (Conversi et al., 2010a; Beaugrand 2015). Möllmann and Diekmann, (2012) suggested that multiple drivers, such as climate and overfishing, may interact in triggering ecosystem community shifts between alternative states. Reid and Beaugrand (2012) observed that, in many cases, the reported shifts coincided with major temporal changes seen in marine temperature anomalies. The interaction between climate-induced environmental changes and species’ ecological niches (Beaugrand 2015; Beaugrand et al., 2019) may lead to a community shift. For such shifts, the existence of tipping is not needed as an explanation.

Another region of potential climate change-induced regime shifts is the Arctic Ocean. As summer sea ice declines, spring phytoplankton blooms are becoming possible, leading to Arctic ecosystems becoming more like the present North Atlantic and productivity increasing by 30-50 per cent (Yool et al., 2015). Warming and circulation changes can also lead to the spread of invasive species – for example in the Barents Sea and from the Pacific (Kelly et al., 2020; Neukermans et al., 2018; Oziel et al., 2020) (see 1.4.2.1). However, while these changes may trigger regime shifts, it is currently difficult to predict whether they will feature self-sustaining tipping dynamics. 

Empirical thresholds for marine communities have been estimated in specific cases using ecosystem model-derived indicators of community status (e.g. Samhouri et al., 2010), but are in general challenging to identify. Evidence for irreversibility is anecdotal and case-specific. One example is shifts in the anchovy-sardine cycles (Schwartzlose et al., 1999) that occur worldwide. Such shifts appear to be triggered by changes in short and long-term climate conditions. In the Peruvian upwelling system, switches in climate cycles can thus correspond to tipping points for the community (Alheit and Niquen, 2004; Chavez et al,. 2003), with effects on the middle (decadal) to long (centuries) timescale (Salvatteci et al., 2018). Evidence for this system suggests that natural fluctuations and anthropogenic climate change may pose an increased risk of tipping toward irreversible changes to a community characterised by less desirable (from a social-ecological perspective) and less productive features (Salvatteci et al,.2022).

Kelp forests

Kelp forests are mostly coastal ecosystems dominated by dense populations of large brown macroalgae (Figure 1.3.27). In recent decades, a significant number of these forests have undergone devastating collapses, resulting in their transformation into desolate and unproductive communities, called barrens. These collapses are primarily driven by overgrazing by sea urchins (Ling et al., 2015). However, additional pressures, such as marine heatwaves (McPherson et al., 2021), nutrient concentration (Boada et al., 2017) and sedimentation (Foster and Schiel, 2010), also contribute to its formation.

Persistent, catastrophic regime shifts in coastal rocky communities transitioning between productive macroalgal beds and impoverished sea urchin barrens have been shown to occur worldwide (Ling et al., 2015). In many cases, such regime shifts exhibit nonlinear dynamics with hysteresis, where the transition shifts exhibit tipping points (Filbee-Dexter and Scheibling, 2018). Thresholds can be estimated empirically through a critical density of sea urchins (Ling et al., 2015), but such thresholds are influenced by biotic and abiotic factors.

Two feedbacks promote the stability of the barren state: processes that reduce kelp recruitment on barrens and processes that allow sea urchins to maintain high densities on barrens (Filbee-Dexter and Scheibling, 2018). For example, adult sea urchins seem to provide shelter and facilitate survival of urchin recruitment, offering a reinforcing mechanism. Similarly, barren conditions are kept open by intense grazing, reducing the chances of kelp recruitment.

Empirical studies have demonstrated the possibility of kelp forest recovery once sea urchin densities are limited (Smith and Tinker, 2022; Galloway et al., 2023). However, such recovery is influenced by abiotic factors such as marine heat waves, making kelp forest reversibility uncertain.

Biological carbon pump

The biological carbon pump (BCP) refers to the suite of processes that remove ~50 Gt of carbon annually from the atmosphere and into marine biomass, transferring ~10 per cent of this into the deep ocean (Carr et al., 2006; Westberry et al., 2008; Fu et al., 2016). Without this flux, atmospheric CO₂ would likely be ~200 ppm higher than the present-day concentration (Henson et al., 2022). 

The largest component of the BCP, the gravitational pump, is driven by sinking of organic matter, mostly from dead plankton and detritus such as faecal pellets (Figure 1.3.28) (Nowicki et al., 2022). This part of the BCP is expected to decline with warming as a result of reduced mixing between warming surface and colder deep waters (thermal stratification) leading to reduced nutrient supplies for surface algae (i.e. phytoplankton), as well as warming favouring smaller plankton species that contribute less sinking matter (Armstrong McKay et al., 2021). However, there is no known mechanism that would enable this decline to become self-sustaining, with changes scaling quasi-linearly with emissions in models, and it is therefore not considered to show tipping-point behaviour (Armstrong McKay et al., 2022).

A system that is more likely to show tipping-point behaviour is the seasonal lipid (fat) pump (SLP) (Jonasdottir et al., 2015). The SLP mainly occurs in high latitude oceans and is driven by the seasonal vertical migration of lipid-rich zooplankton (Figure 1.3.29) into the deep ocean, where they overwinter for ≥6 months, directly injecting carbon below the winter mixed layer. A dramatic reduction in primary production via diatoms, for example, driven by changing nutrient supply patterns via increased stratification due to ocean warming, could result in zooplankton not consuming enough lipids to successfully overwinter and reproduce the following spring. Arresting the SLP would irreversibly change the ecological and biogeochemical functioning of high latitude ecosystems.

Deep ocean warming will increase rates of respiration, meaning that lipid reserves may become exhausted before returning to the surface. This will interrupt recruitment and halt the SLP. The poleward migration of non-diapausing species (i.e. those that do not form an inactive life-form for parts of the year), as polar conditions ameliorate, could eventually mean that lipid-storing deep-diapausing zooplankton eventually disappear and the SLP collapse will be irreversible. However, the SLP was only described <10 years ago, and so our nascent understanding of its scale and complexity currently precludes the establishment of thresholds.

Other parts of the ocean biological pump could also result in nonlinear dynamics or tipping points. A recent paper found evidence that ‘mixotrophs’ – plankton that can both photosynthesise like algae and consume other plankton – can switch between a photosynthesis-dominant carbon sink state to a consumption-dominant carbon source state, with warming pushing them towards the latter and nutrient pollution making tipping dynamics more likely (Wieczynski et al., 2023). Mixotrophs are common in the ocean but their role in ocean and ecosystems and the biological pump is under-studied (Ward, 2019), making the impacts of these potential tipping dynamics unclear.

Marine oxygenation

Coastal hypoxia is a regime shift that occurs when dissolved oxygen in water diminishes below levels detrimental to marine life. As a consequence, one of their symptoms are ‘dead zones’, areas of the oceans where fish and many other marine organisms (particularly in benthic communities) migrate outwards or die due to low oxygen levels.

While hypoxia is naturally occurring in some areas, hypoxic events have been increasing over the last few decades. A first global assessment of ocean deoxygenation documented over 300 cases mainly in the Atlantic coast of North America, the Caribbean, Mediterranean and Baltic seas (Diaz and Rosenberg, 2008). Subsequent assessments expanded to >500 case studies, from occasional hypoxic events to severe anoxia (Breitburg et al., 2018) (Figure 1.3.30).

The main mechanisms underlying coastal hypoxia are related to over-enrichment of nutrients like phosphorus and nitrogen coming from agricultural fertilisers, sewage or upwelling currents in the ocean. The latter are natural currents that bring nutrient-rich waters from the deep ocean to the surface, powering the primary producers (i.e. algae) and in turn productive food webs. In high-nutrient waters, algae can become over-abundant, consuming the available oxygen and causing the death of fish or other oxygen-dependent organisms. As they die, decomposers then further decrease available oxygen as they break down extra organic matter. Additional nutrients from fertiliser and sewage runoff on land is amplifying this process, increasing the number of hypoxic events and sites (Breitburg et al., 2018; Heinze et al., 2020).

At the same time, phosphorus can be released from sediment under low oxygen conditions, acting as a positive/amplifying feedback by further amplifying the growth of algae and the consumption of oxygen (Conley et al., 2002; Adhikari et al., 2015). Besides nutrients, climate change can exacerbate hypoxia by reducing oxygen solubility in water (Breitburg et al., 2018), and is projected to cause widespread deoxygenation over coming centuries to millennia via warming and enhanced land weathering delivering more phosphorus (Watson et al,.2017; Battaglia and Joos, 2018). Even if warming peaks and falls (‘overshoot’) Earth system models indicate that deoxygenation in the upper 1000 metres of the ocean is irreversible for multiple centuries (Santana-Falcon et al., 2023). Sea surface temperature can also change the strength of upwellings and thus the inflow of nutrients in coastal ecosystems.

Marine ecosystems with dissolved oxygen higher than >2 mL per litre sustain diverse ecological communities, and this level is considered normal (also known as ‘normoxia’). Below this level the symptoms of hypoxia appear, including hypoxic events and dead zones. Anoxia occurs when levels of dissolved oxygen are below 0.5 mL per litre, which only a few microbial species are able to survive (Diaz and Rosenberg, 2008). Some dead zones and hypoxic events are reversible in scale of months to years. However, more and more areas are reported as chronically hypoxic, possibly irreversible in the timescale of ecosystem managers (centuries). Examples of severe hypoxia are dead zones in the Gulf of Mexico, central Baltic, Kattegat, Black Sea, and East China Sea (Breitburg et al., 2018).

Assessment and knowledge gaps

Table 1.3.3: Summary table of marine environment tipping points considered in this section.

System	Tipping system?	Timescale	Biophysical Impacts	Confidence	Gaps
Fisheries[Small, fast-growing fish]	No	Decades	Changes in entire trophic assemblage. A regime shift in one species could propagate the regime shift in many components of the ecosystem. Important especially in bottom-up and wasp-waist ecosystems.	Low confidence because too many different species	Need more coherent statistical approaches to identify tipping points and the presence of hysteresis. Also need more analyses on single species that look at tipping points in fisheries.
Fisheries [Large, slow-growing fish]	Depends on the stock	Decades	Changes in entire trophic assemblage. A regime shift in one species could propagate the regime shift in many components of the ecosystem. Especially important in top-down ecosystems.	Low confidence because too many different species and many different areas	Need more coherent statistical approaches to identify the tipping points and the presence of hysteresis. Also need more analyses on single species that look at tipping points in fisheries.
Fisheries [Cod]	Yes (in 16 out of 19 stocks)	Decades	Changes in the entire trophic assemblage, trophic cascade.	High confidence	In some cases there is the need to better understand feedbacks of hysteresis.
Community shifts	Yes	Decades	Changes in ecosystem function. structure and feedbacks that may affect how to best manage the system.	Low confidence – complexity from many different species and interacting drivers	Better understanding required on interplay of multiple drivers and species interactions. Tipping points difficult to identify and predict.
Kelp forests	Yes	Months to decades	Changes to community composition of fish and macroinvertebrates scaling up to trophic disassembly.	High confidence	Necessary to understand how key ecosystem properties, e.g. resilience or stability of kelp forests, evolve over the years.
Ocean hypoxia	Yes	Months/years to centuries. Reversible at surface, irreversible at depth for centuries to millennia	Major changes in ocean productivity, biodiversity and biogeochemical cycles.	Low confidence	Degree of self-sustaining change and hysteresis; influence of future climate change and nutrient use.
Biological pump [Seasonal Lipid Pump]	Potential, but uncertain	Decades	Major changes in trophic transfer, carbon sequestration and ocean biogeochemistry.	Low confidence	Better understanding of zooplankton physiology and its response to environmental change.

Table 1.3.3 summarises our assessment of tipping dynamics (with confidence levels) along with biophysical impacts, timescales and knowledge gaps for marine ecosystems. We have high confidence that cod fisheries and kelp forests can pass tipping points, low confidence that some other large-fish fisheries, marine communities and potentially the lipid pump could also tip, and medium confidence that marine hypoxia could feature tipping dynamics. Knowledge gaps include limited understanding of complex species and driver interactions, limited ability to detect and project marine tipping points in practice, and how ecosystem resilience can change over time.