Regime Shifts DataBase

Large persistent changes in ecosystem services

Steppe

The steppe vegetation coverage is dominated by grasses growing in dry soils (Eroglu et al. 2012). In this regime, herbivores increase nutrient availability by accelerating nutrient cycling. This increases primary production. During the Pleistocene epoch, arctic steppes sustained large grazing herbivores such as mammoths, bison and yaks (Zimov 2005). Today, grasses still sustain reindeer/caribou population. Currently, the steppe communities of Yakutia, Siberia are confined to the terraces above floodplains and south facing slopes of the river valleys (Yurtzev 1982).

Tundra

The tundra vegetation coverage is dominated by mosses and shrub such as dwarf birches and willows growing in waterlogged soils. The ecosystem has lower primary production and nutrient cycling thus the growth rate is slow. Moss growth is limited by moisture so the moist and nutrient low soil promote and sustain moss growth.

Herbivory has caused large-scale vegetation changes across the Arctic (Speed et al. 2009; Bråthen et al. 2007; van der Wal 2005; Srivastava & Jefferies 1996). Herbivores trample the vegetation and deposit faeces to nutrient limited soils which promotes grass growth sustaining herbivore population. In the absence of herbivores grass litter increases causing soil moisture to increase which reduces nutrient availability promoting the growth of unpalatable mosses.

Climate can act as a driver by changing air temperatures which is associated with changes in soil moisture. Soil moisture is key component in sustaining moss or grass coverage. Moss growth is more limited by water than by nutrients. However, nutrients are more available in drier soils which promote grass growth. Currently, climate is changing, in the long term whether steppe, tundra or another regime will dominate Arctic ecosystems in uncertain.

Shift from Steppe to Tundra
Grass-dominated steppe is maintained by trampling and nutrient inputs from herbivore faeces. The enriched soils promote the growth of grass coverage (Shaver et al. 1986). Grasses have a strong influence on soil moisture as they dry out the soil through high transpiration (Zimov 2005). Dry soil conditions are associated with increased nutrient availability intensifying grass growth (Nadelhoffer et al. 1991).

In the absence of herbivores, moss coverage increases creating water-logged soils (Zimov 1995). Water-logged soils prevent decomposition of organic matter, so nutrient availability decreases, which limits the growth of grasses and, hence, promotes moss growth instead (Zimov 1995). As moss growth increases, herbivore density is further reduced because moss have lower nutritional value.

Moss-dominated tundra is susceptible to trampling, a disturbance caused by large herbivores; therefore, moss coverage increases with low herbivore density (van der Wal 2001). Mosses causes an increase in soil moisture. The combination of high soil moisture creates the conditions for water-logged soils establishing optimal conditions for moss growth (Zimov 1995).

Shift from Tundra to Steppe

Climate is the main driver that could shift tundra back to steppe; this is because large grazers don't exist any more and arctic ecosystems are vulnerable to climate change. Rising of air temperatures would deepen the thawing layer of permafrost and increase evaporation. Therefore, frozen nutrients from soil would become available, which would promote grass growth; in addition to that, higher temperatures will enhance organic matter decomposition and hence nutrient availability. Then, grasses will increase evaporation. Nevertheless, moist soil conditions might be resilient since permafrost melting will make frozen water available. If there are no draining conditions, moisture will provide a suitable environment for mosses that will insulate soil temperatures and then buffer permafrost melting.

Rising of temperatures will create dynamics on arctic ecosystems that are difficult to predict. However, recent evidence shows that tundra could turn into different types of vegetation (e.g. shrub land, forest, lake, grassland) depending on local conditions (Karlsson et al. 2011). Steppe-grasses will prevail on drier areas, but more probably as patches within a mosaic since large herbivores that maintained steppe conditions during the Pleistocene, are not present.

Shift from Steppe to tundra

Shifting from steppe to tundra had a great impact on provision of food. The thermal insulation of the tundra prevents permafrost melting, which prevents destabilization and collapse of infrastructure (Schaefer 2012). Increasing air temperature due to climate change deepens the active layer of permafrost. However, soils covered by moss, permafrost is less susceptible to degradation (Ivanova 2003), this ecosystem service is essential since permafrost contains almost twice as much carbon as the atmosphere today and its effects would be irreversible at human time scale (Schaefer et al. 2012).

The wellbeing of nomadic indigenous peoples such as Nenets, Enets, Sami, Nganasans and Selkups, are connected to Arctic that provide opportunities for grazing and food sources, as well as ecosystems that co-produce cultural ecosystem services i.e. production and conservation of indigenous knowledge, practices and believes, as well as production of scientific knowledge (Huntington 2013). Tourism is also a cultural ecosystem service that will be affected.

Tundra to steppe

Arctic soils covered by mosses, permafrost is less susceptible to degradation (Ivanova 2003). Transition from tundra to steppe implies loss of this buffering layer. Consequence, the following ecosystem services will be impacted: Water availability. Local and regional freshwater dynamics will change leading to a decline in the number of lakes and wetlands (Wrona et al. 2016; Schaefer 2012). Increased abundance of grassland will affect livestock production, steppe might benefit both reindeer hunters and herders. Protection from infrastructure destabilization; pipelines, railways and power lines across the arctic are built on solid ground provided by the permafrost that if degraded their stability might be lost. This might cause economic costs and serious ecological damages (Schaefer, 2012). Fire regulation; as moss tundra shifts to a drier regime, possibilities of wild fires will increase, which contributes to thermal erosion of permafrost (Schaefer, 2012). Prevention of climate change; permafrost degradation turns permafrost's carbon stock into a carbon source, which would accelerate climate change (Schuur et al. 2015). Loss of this ecosystem service impacts the global level, since permafrost contains twice as much carbon as the atmosphere and effects of release is irreversible at human time scale (Schaefer, 2012).

Cultural ecosystem services will also be affected; conservation of indigenous knowledge, practices and believes related to tundra would be lost, affecting the well being of t he Nenets, Enets, Sami, Nganasans and Selkups (Huntington 2013).

To prevent the steppe from shifting into tundra, greenhouse gas emissions must be reduced in order to avoid the effects of climate change. The planetary boundary value for atmospheric CO2 concentration is 350 ppm above pre-industrial level (Rockstrom et al. 2009). This was further supported by the COP21 agreement, a treaty signed to limit temperature increase to 1.5^oC compared to pre-industrial levels. These efforts are not enough to prevent the feedback loop, eliminating emission is the only solution. In the Artic, land management strategies can address feedback loop on the local level. In a study, by Post and Pederson (2008), a 5-year experimental investigation of Arctic plant community response to warming, showed that warming under continuous grazing pressure from muskoxen did not differ from plant composition without warming suggest that management of large herbivores may be an important aspect for mitigating ecosystem response to future climate change

Attempts are being made today to restore the Pleistocene steppe ecosystem on a limited spatial extent in Siberia. Sergey Zimov's Pleistonce Park, 160 km², attempts to reestablish the steppe ecosystem by reintroducing megafauna to Northern Siberia (Republic of Yakutia). Zimov (2005) and colleagues tried to test if the 'key-herbivore´ hypothesis can be verified, as today's Holocene climate should be optimal for the Pleistocene steppe vegetation. The megafauna consisting of reindeer, moose, Yakutian horses, musk oxen and bison would influence the vegetation and soil composition, by trampling on grassland and returning nutrients to the soil through their manure (Zimov 2005). The grass root systems stabilize the soil and trampling reduces the albedo, exposing ground to colder temperatures, both would prevent permafrost from melting.

Due to thermokarst processes such as permafrost melting, numerous areas of the Arctic landscape are extremely lake-rich.  The water balance of those lakes may either be maintained due to impermeable permafrost that limits groundwater flow (Marsh 2009) or lost due to seasonal drainage, a natural characteristic of the Arctic system. However, a widespread, unnatural decline in lake abundance is being observed within the Arctic due to decades of rising atmospheric temperatures in the region that has been deteriorating the state of the permafrost (Smith et al. 2005). The lake-rich landscape of the Arctic is therefore slowly transforming into a drained, dry landscape increasingly dominated by vegetation such as shrubs and graminoids (Hinzman 2005).

Thermokarst lake ecosystem

The thermokarst lake-dominated landscape is formed as a response to some imposed disturbance, such as road construction, wildfire or climatic warming, which causes permafrost to thaw, creating irregular surface topography depressions, called thermokarsts. Those may appear as near-surface massive ice melts, allowing the surface to subside (Hinzman et al. 2005). As permafrost melts over time, those depressions become filled with melted water forming the lakes characteristic of the Arctic region.

Terrestrial ecosystem

Driven by external and internal factors, Arctic lakes are drying up at an unnatural speed within the region. This shift is accompanied by changes in vegetation, for example, increased shrub- and graminord abundance and a decrease in the cover of mosses and lichens through a broad expanse of the Arctic (Hinzman et al. 2005). The changes in vegetation may then further transform the landscape through sequential effects on foraging mammals and birds, as well as aquatic fauna and insects (Hinzman et al 2005).

Shift from thermokarst lakes to terrestrial ecosystem

The main external driver of this regime shift is climate change, which is caused by elevated greenhouse gas emissions (IPCC 2007). Increased temperatures warm the soil and warms the permafrost. As the permafrost thaws, greenhouse gasses such as CO² and CH₄ that are trapped in the ice are released into the atmosphere, further increasing the effects of climate change (Hinzman 2005; Karlsson et al. 2011). At a local scale, the effects of climate change are incremental. However, as this interaction is occurs over large areas, the effects are aggregated and self-reinforcing, which is significant since permafrost and seasonally frozen grounds store about 25% of the total global soil carbon stock (IPCC 2007).

At a local scale, initial thawing of the permafrost from increased atmospheric temperatures creates a dynamic landscape of depressions (thermokarst) in which lakes are formed (IPCC 2007). As permafrost become more degraded, drainage through outlet channels occurs. Since the temperature of melted water is higher than that of ice, it will exacerbate the melting of permafrost, causing another reinforcing feedback. Once the layer of frozen ground is completely penetrated the lakes can become permanently drained allowing for vegetation to be established (Marsh et al. 2008).

Shift from terrestrial ecosystem to thermokarst lake ecosystem

Heightened air- and soil temperatures melt the ice-rich soils and permafrost on which forests in the Arctic region develop, changing the physical conditions for the forest growing on top of it. When the roots of the trees get flooded, the trees die and ponds and lakes eventually replace the forest (Hinzman 2005). However, it is important to note that the shift from a terrestrial ecosystem to a lake dominated one is not as frequent as the shift from lakes to terrestrial ecosystems. The reason for this are the reinforcing feedbacks as mentioned above.

The warming of soil temperatures leads to the melting of continuous permafrost and the subsequent increase in abundance of discontinuous or sporadic permafrost. Melting of permafrost causes depressions in the soil called thermokarsts. Increased precipitation in the form of snowfall may have either an insulating or cooling effect on soil temperatures depending on several variables such a timing, duration, accumulation, and melting processes of seasonal snow cover, density, structure, and thickness of seasonal snow cover etc. (Zhang 2005). Depending upon the local surface energy balance, the thawed ground may refreeze or the permafrost can continue to degrade (Hinzman et al. 2005). On sites where continuous permafrost is thawing the active layer of soil in which melted water can move deepens allowing for the formation of thermokarst lakes (Hinzman et al. 2005). In freezing seasons, if the temperature is low enough and the hydraulic parameters on site suitable, the formation of these lakes can contribute to the re-formation of permafrost soils. Ice-rich permafrost can also prevent surface water from infiltrating to deeper groundwater zones, causing surface soils to be very wet. Decreased albedo due to changes in vegetation, the extension of snow-free and ice-free periods on terrestrial and lake surfaces and abundance, and reduction in the area occupied by glaciers and continental ice sheets in high latitudes may act as a positive feedback into atmospheric warming (Hinzman et al. 2005).

On sites where there is an abundance of discontinuous permafrost the active layer also deepens allowing for increased surface area for sub-ground water movement that increases the rate of drainage. Thermokarst lakes and ponds may begin to fill or drain depending upon the direction of the hydraulic gradient beneath the lake (Hinzman et al. 2005). Therefore it is not the depth of the active layer alone that determines the rate of drainage. However, increased drainage of lakes may prevent the reformation of permafrost from lack of available water for freezing in winter months and it can be inferred through advection of heat from decreased albedo. Once these drained lake beds become permanently drained they are likely to be re-vegetated which will change the albedo and may cause the extension of snow-free and ice-free periods and may act as feedback into increasing atmospheric temperature (Hinzman et al. 2005).

The affects of the shifts from lake to terrestrial ecosystems are aggregated at a local, regional and global scale due to the large scale and frequency at which those dynamics occur. The whole regime as described in the causal loop diagram below is a re-enforcing feedback that at its current state is leading to an overall decrease in lake dominated ecosystems and a increase in terrestrial ecosystems (Smith et al. 2005).

Even though this regime shift occurs at the local scale, its consequences go beyond that scale because it reshapes regional hydrology (through changes in water balance and surface water- connection and fragmentation) and alters regional supplies of the ecosystem (Karlsson et al. 2011). This regime shift therefore affects provisioning ecosystem services, as it will decrease the availability of freshwater. The above-ground stored freshwater, which is of vital importance to migratory birds, fish, and other wildlife used by indigenous people, will shift to below-ground stored freshwater (Artic Science Journeys Radio Stories 2005). Globally, however, freshwater will become a more abundant ES due to reductions permafrost and subsequently ice sheets and glaciers resulting from the regime’s feedbacks to climate change. Nevertheless, this global increase in the freshwater ES is not a positive one: “Increased freshwater delivery to the Arctic Ocean from reductions in ice sheets and glaciers result in rising sea level (...)” (Hinzman et al. 2005). At a local scale traditional hunting and fishing practices will be impaired (Vincent et al. 2013).  Arctic freshwater systems provide important migratory routes for fish stock. Due to this regime shit, those routes would be greatly altered in connectivity among lakes and river channels, as well as in terms of their physical coupling to the coastal marine ecosystem (Vincent et al. 2012). Globally, fishing practices would suffer as well. Increased discharge of water into the Arctic Ocean due to draining results in increased nutrient and external organic matter inputs to the Arctic Ocean that affect primary processes at the base of the marine food web (Vincent et al. 2012). Climate regulation is heavily affected on a global scale through the processes of this regime shift: as it sets up a positive feedback cycle in which the release of methane and carbon dioxide through the newly exposed soil and melted permafrost feeds into global warming. The shift from lakes to terrestrial ecosystems heightens gully erosion, a type of erosion that occurs when water is channeled across unprotected land and washes away the soil along the drainage lines. As permafrost degrades further, it causes erosion and slumping of lake edges and stream channels

The Arctic lakes are of vital importance to indigenous people as they are a source of fresh drinking water and also provides important winter transport routes. These lakes not only allows indigenous people access to their traditional hunting and fishing areas, but also to transport goods to remote communities and industries such as mining centers (Vincent et al. 2013). The regime shift would also mean an increase in products extracted from woods such as timber commodities and fuel: “increased areas of tree growth in the Arctic could serve to (…) take supply more wood products and related employment, providing local and global economic benefits” (Hassol et al. 2004).

There are few management options available to maintain the thermokarst lake ecosystems due to the long time scale at which permafrost melts and the natural dynamics within this system e.g. hydraulic fluctuations or site-specific soil conditions. One management option is to continue to reduce the emissions of greenhouse gases while simultaneously investigating further opportunities for carbon offsetting schemes.

For hydroelectric reservoirs, shifting ice conditions will have both positive and negative effects, and may require adaptive changes in operating procedures, with attention to minimize negative impacts associated with ice jams and ice breakup downstream of the spillway (Vincent et al. 2013). Fisheries management plans will also need to be adapted to the changes in migration and productivity of northern fish populations with ongoing climate change (Vincent et al. 2013). These essential resources require the development of integrated freshwater management plans, which include consideration of alternate water sources as traditional supplies change in quantity or quality (Vincent et al. 2011). The Arctic will require increasing vigilance and appropriate water management strategies to avoid and minimize the impacts of changing water impacts in the future (Vincent et al. 2013).

The boreal forest in northern Interior Alaska covers an approximate area of 48 million hectares (Mann et al. 2012). Bounded by the Alaska Range to the south, the Seward Peninsula to the west and Brooks Range to the north (Wolken et al. 2011). It is characterized by large areas of gently sloping uplands and flat lowlands isolated by mountain ranges and braided rivers with broad floodplains. During the past 50 years, the mean annual temperature in the interior boreal forest of Alaska has increased by 1.3 ºC (Hartmann and Wendler, 2005) and this has been strongly associated with recent increase in fire frequency and severity (Wolken et al. 2011). The annual area burned in Interior Alaska has doubled during the last decade (Kasischke et al. 2010). The forest has been dominated by coniferous trees (mainly black spruce), which is a group of species that thrive under colder, moist conditions in a deep soil organic layer. However, due to climate warming, Interior Alaska is currently shifting towards a deciduous dominated forest (e.g. aspen and birch) that prefer warmer and drier conditions with a shallow soil organic layer (Johnstone et al. 2010).  

Low fire frequency coniferous dominated forest

Under cold and wet conditions, coniferous trees, such as the black spruce (Picea mariana), dominate the boreal forest in Interior Alaska. The coniferous trees promote the accumulation of a deep soil organic layer, cold and moist soil temperatures, and slow decomposition rates. Due to the moistness of the soil, fires are not frequent in these forests. However, when they occur, they tend to be severe due to the high flammability of the black spruce. Due to the high water content of the soil organic layer, it will only be entirely consumed by the most severe fires (Johnstone et al. 2010). As long as the entire organic layer has not been consumed, the coniferous trees have a regenerative advantage due to their seed’s ability to sprout in moist organic soils. This means that these forests tend to return to a coniferous dominated state even after a fire (Hollingsworth et al. 2013).

High fire frequency deciduous dominated forest

A deciduous dominated forest, mainly consisting of aspen and birch trees in Interior Alaska, prefers and reinforces dry and warm soil conditions. Although these forests have a high organic litter production, the fast decomposition rate and the subsequent fast nutrient turnover, leaves a shallow soil organic layer and a nutrient rich mineral soil that favours deciduous tree regeneration. The warm and dry conditions in the forest lead to a comparatively high fire frequency, but due to the low flammability of the deciduous trees, the fires rarely become very severe (Johnstone et al. 2010). However, because of the shallow pre-fire soil organic layer, the fires tend to expose the mineral soil, which favours the post-fire reproduction of deciduous trees (Hollingsworth et al. 2013).

The main external direct driver of the regime shift is climate warming, which affects the important feedbacks of the boreal forest ecosystem. Warmer and drier climatic conditions have a direct effect on the most important soil characteristics (i.e. soil temperature, moisture and depth of the soil organic layer), influencing the vegetation cover. Coniferous trees require a cold moist soil with a deep soil organic layer to successfully reproduce. Deciduous trees, on the other hand, are competitively superior under warmer and drier soil conditions with a shallow soil organic layer (Johnstone et al. 2010). Extended summer seasons and the accompanying drier conditions create a suitable environment for an increase in the annual number of natural ignitions by lightning strikes (Kasischke et al. in press). Fires have an immediate effect on the environmental conditions of the forest, affecting factors such as soil moisture and the post-fire forest community composition (Kelly et al. 2012). Once established, the trees support and emphasize the conditions that favour their own reproduction and dominance. Under colder environmental conditions, coniferous trees, such as the black spruce, produce large amounts of organic litter, which, together with the presence of permafrost, maintains a moist soil and slow decomposition rates, contributing to a deep soil organic layer. Under warmer climatic conditions, deciduous trees produce ground layer fuels that decompose at fast rates, maintaining a shallow organic layer (Johnstone et al. 2010). Global socio-economic development indirectly drives climate warming, contributing to the increased atmospheric temperatures through e.g. greenhouse gas emissions (IPCC, 2011). Fire management can have a limited effect on the regime shift through fire suppression practices.

Shift from coniferous to deciduous-dominated forest

The coniferous dominated boreal forest of Interior Alaska requires a deep soil organic layer to succeed over other classes of trees. Once the required depth is present, this has a strong reinforcing effect on the coniferous forest. This is a requisite to maintain cold moist soil conditions in which coniferous trees thrive. The cold soil temperature reduces the decomposition rate of organic litter, maintaining a deep soil organic layer (Johnstone et al. 2010).

External shocks such as severe and frequent fires or insect outbreaks might cause dramatic shifts in the soil and vegetation by impacting the soil organic layer depth and tree succession (Hollingsworth et al. 2013; McCullough et al. 1998; Wolken et al. 2011). If the severity of such external shocks is sufficient, thresholds in the most important regime characteristics will be crossed (e.g. shallow post-shock soil organic layer) creating windows of opportunity for deciduous species to dominate. Atmospheric temperature influences the frequency and severity of fires, the soil characteristics and the length of the growing season; therefore, climate warming can also cause a change in conditions that would set the grounds for a shift to a new stable state (Mann et al. 2012).

The deciduous dominated forest has a competitive advantage in warm and dry conditions with shallow soil organic layers. Again, once the required shallow layer is present, it has a strong reinforcing effect on the deciduous forest. A shallow soil organic layer maintains warm and dry soil conditions that speed up the decomposition of organic litter, maintaining the required depth of soil organic layer (Johnstone et al. 2012).

Shift from deciduous-dominated  to coniferous forest 

In order to shift from regime 2 to regime 1, a prolonged climatic cooling would be necessary. A slow change of the environmental conditions towards those that favour a coniferous boreal forest could, if the forest is left undisturbed for a sufficiently long time, lead to late successional coniferous species, like the black spruce, replacing early successional deciduous species (Rupp et al. 2002). Although the spatial scale considered is of extreme importance, sufficiently large external shocks, such as deciduous-specific insect outbreaks, could clear sufficiently large areas to advance the dominance of coniferous trees. In such hypothetical events, coniferous trees could get the opportunity to sprout and start altering the soil conditions to their own favour.

The provisioning of wild products such as berries and game is expected to shift some species and decrease others, but it is unclear what the net changes will be (Chapin et al. 2008). The boreal forest is likely to become a net source of carbon to the atmosphere (McGuire et al. 2009; Schuur et al. 2009), while the air quality is expected to decrease periodically due to the increase in fire frequency (Chapin et al. 2008).

The primary tactic used for reducing impacts of fire and enhance resilience (i.e. maintain structures, feedbacks, functions and processes of regime 1 is fire suppression to reduce the fire severity (Chapin et al. 2008). Decrease in fire severity could prevent the fire from consuming the soil organic layer, which in turn favours the conditions for coniferous tree. Fire suppression should be maintained in Alaska since it minimizes the risk fire imposes to life and property in communities. However, distant fires should be allowed to burn under controlled conditions, since it could have a cooling effect at a regional scale due to the higher albedo of post-fire deciduous species compared to the previous coniferous forest (Chapin et al. 2008).

Reducing air temperature is essential to be able to restore regime 1. Alaska itself accounts for a miniscule proportion of global greenhouse gas emissions and consequently to climate warming (Chapin et al. 2008). Therefore, reducing Alaska’s emissions will not have a major impact on global climate change. Global abatement of emissions is the long-term solution. Regardless of global greenhouse gas emission policies, it is highly likely that recent warming and wildfire frequency will continue for several decades in Alaska due to the multidecadal lag in the climate system (IPCC, 2007). This makes the reduction of emissions an inefficient short-term solution.  Secondary challenges that face efforts to achieve reduced emissions include restricting the economy and changing the focus of developed nations on a continued economic but sustainable growth. Educating global public about the social impacts of climate change in Alaska and other parts of the world (e.g. reduction in ecosystem services) may promote the willingness to reduce emissions (Chapin et al. 2008).

In summary, this means that the regime shift from coniferous to deciduous dominated forest in theory could be reversible through manipulation of e.g. soil conditions or local climate. However, due to projected global socio-economic development and climate change, it is highly unlikely that reversing the regime shift will be possible during the next 100 years.

This regime shift involves an abrupt reorganization of an aquatic food web due to a decrease in top predators or increase in environmental factors like temperature. Trophic cascades play a central role in these regime shifts. During these cascades groups in adjacent trophic levels often show inverse patterns in their abundance (Carpenter 2003). In other words, every other trophic level displays an increase in biomass, which leads to increased predation pressure and biomass decrease on every other trophic level. Most trophic cascades are described in ecosystems with low species diversity and/or simple food webs and/or small size (Frank et al. 2005). The magnitude of a trophic cascade varies depending on species diversity, regional oceanography, local physical disturbance, habitat complexity and fishery practices (Salomon et al. 2010). Food webs with more trophic levels and biodiversity show less trophic cascades than simpler systems (Salomon et al. 2010); due in part to omnivory and the variation on trophic interaction strength (Bascompte et al. 2005). However some documented examples includes subtidal reefs in New Zealand (Shears and Babcock 2002), Caribbean coral reefs (Bellwood et al. 2004), the Gulf of Maine (Steneck et al. 2004), and the North Atlantic (Kirby et al. 2009).

Predator-dominated food web.

Ecosystems in this regime are characterized by high predatory fish, low planktivorous fish, high zooplankton and low phytoplankton abundance. Such ecosystems can be subject to human influence, such as intensive fishery, but exhibit so called “natural compensation” e.g. via species richness and omnivory (Pace et al. 1999) that maintain the regime. Despite high productivity, such ecosystems are less likely to suffer from eutrophication due to high grazing pressure on phytoplankton. Ecosystem services associated with predator-dominated food webs includes food provision, higher biodiversity, better disease and pest control, as well as cultural services like recreation and aesthetic values (e.g. diving, sport fishing).

Planktivore-dominated food web.

Ecosystems in this regime are characterized by low predatory fish, high planktivorous fish, low zooplankton and high phytoplankton abundance. In extreme cases the actual number of trophic levels (TLs) can decrease. Ecosystems with lower trophic level dominance are more vulnerable to eutrophication due to lowered grazing pressure on phytoplankton, as well as invasion of planktivore organisms like jellyfish. A planktivore-dominated food web can affect the fluxes of carbon, transforming aquatic systems -coastal and lakes- from sinks to sources of green house gases (Bakun et al. 2010; Estes et al. 2011). Other ecosystem services like fisheries and recreation are expected to be significantly reduced.

Trophic cascades are usually initiated by fishing pressure that have direct or indirect impacts on the abundance of top predators. The response time to these drivers varies across ecosystems. In the case of coastal areas it has been found that the ecosystem response time to overfishing can vary from decades to centuries where multiple predators exist (Jackson et al. 2001).

Other factors that affect the regime shift are global warming and demand of food and fiber. Global warming have effects on upwelling, water circulatory systems that bring nutrients from the bottom of the sea towards the surface. By altering upwellings, food webs receive either too much or too little nutrients. In both cases the energy transferred towards higher trophic levels diminish as nutrients input variability increase. Demand of food strongly drives fishing, being the main cause of lost of top predators over the world.

Predator dominated food webs occur in environments with low disturbance both from fishing and nutrients input. They usually have four or more trophic levels. The regime is maintained by biotic mechanism that comprise predation and competition both intra and inter species. Abiotic factors also play a role maintaining the regime, by allowing an intermediate level of nutrient inputs from the deep ocean necessary for sustaining the food web.

In general, a combination of low top predator productivity due to poor recruitment or somatic growth combined with high exploitation pressure are conditions under which regime shifts in aquatic food webs are likely to occur (Daskalov et al. 2007). Overfishing can unbalance trophic relations, changing the configuration of competition and predation. Weak competitors may become stronger given less fishing pressure on such stocks, or by favoring the reproduction of it’s preys by fishing its competitor species. In addition, changes in nutrients inputs can represent an advantage for lower trophic levels, allowing them to escape the control of its predators.

As result, the ecological function of certain species groups is changed, allowing over abundance of immediate lower level. For example, the decrease of top-predators allows over abundance of meso predators that in turn reduces planktivorous fish and increase zooplankton. Further increase in nutrients from upwellings can lock the food web in a state dominated by planktivore fish as jellyfish. Low nutrients input, on the other hand, would reduce the amount of energy available for the food web and as a result an abrupt reduction of higher trophic levels is expected.

Shift from predators to planktivore dominated food webs

Commercial fish stocks that are locked into a regime where they are maintained at low levels of abundance has obvious economic consequences on fisheries industry. However, these fisheries related shifts are not always negative, as for example a decrease in top predators can increase the catches of mid-level prey (e.g. shrimps and clupeids) (Frank et al. 2005). When an ecosystem-wide regime shift occurs it affects not only targeted groups, such as commercial fish (Daskalov et al. 2007) but also impacts other ecosystem functions. Biodiversity as well as ecosystem resilience may be reduced by a shift from an ecosystem with high top predator diversity and dominance to a system with lower trophic level dominance. This means that ecosystem may become more vulnerable to both climatic and anthropogenic change. Such ecosystems may more easily suffer from eutrophication, hypoxia and invasion by non-native species. Eutrophication and hypoxia in turn lead to increased algal blooms and decrease the recreational values of water bodies.

Predators and commercial fish stocks have remained under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the reduction of food webs, which in turn increases the risk of fisheries collapse. Depletion of fish stocks has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Hassan et al. 2005). Fish catches are projected to further decrease in the 21st century affecting protein sources for people, especially in poor regions (Hassan et al. 2005). The estimated current contribution of fisheries to human protein consumption is 29 million tons produced industrially and 24 million tons in small-scale fisheries (Hassan et al. 2005).

Options for enhancing resilience

In food webs that exhibit strong top-down control, decreasing the fishing pressure on top predators is an obvious strategy (Scheffer et al. 2001; Frank et al. 2005). However, due to the hysteresis effect, return to a predator-dominated food web may often require substantially lower fishing levels than that which induced the regime shift, together with additional measures. Hence, restoration is usually costly, especially to fisherman (Moellmann et al. 2009). As mentioned above, environmental factors can play an important role in maintaining alternative food web regimes. Nutrient loading is one of the key parameters controlling processes such as eutrophication and hypoxia. However, decreasing nutrient loads in marine environments requires large scale actions and often collaboration at international level (e.g. Helsinki Commission Baltic Sea Action Plan in the Baltic Sea). Hence, the time-frame of these actions in addition to time-lag in ecosystem response is often long. Climatic variation cannot be directly controlled by human actions. However, climate impacts can be mediated by employing other management measures that contribute to maintaining healthy marine ecosystems with high biodiversity and resilience, which then can compensate for changes in climate.

Options for reducing resilience to encourage restoration or transformation

In addition, in small closed food webs systems (3 to 4 trophic levels), biomanipulation has been suggested to manage one of the symptoms of degraded marine environments, namely eutrophication. This involves increasing the population of predatory fish such as bass, pike and walleye through stocking or reduced angling quotas. Increased populations of these predators leads to a decrease in the level of zooplanktivores. This in turn allows an increase in the population of planktivores that graze on the algae and zooplanckton, helping to reduce the algal density and phytoplankton respectively(Smith and Schindler 2009). This option, however, has limited reach when it comes to open marine foodwebs, where seasonal migrations and metapopulation dynamics are common.

During the 20th century global fisheries greatly expanded in terms of effort and range. This expansion resulted in the overfishing of many fish stocks, and in some cases triggered fisheries collapses, for example the collapse of the highly productive Newfoundland cod fishery (Worm et al. 2009). While overfishing can severely deplete fish stocks, it will not necessarily trigger a fisheries collapse. A collapse occurs when once fishery pressure has been removed, stocks fail to rebuild (Hutchings 2000, Kirby et al. 2009), suggesting that the system has become locked into an alternate regime. In this case, alternate regimes are usefully conceptualized as alternative food web structures:

High abundance of a commercial fish species.

In this regime a valuable fish species is common and the fishery is highly productive. The main ecosystem services associated with high abundance of commercial fish are the production of food in first place, employment for both artisanal and industrial fishermen; followed by pest and disease regulation by healthy food webs, recreational values as recreational fishing, knowledge and educational values, as well as maintenance of biodiversity.

Low abundance of the commercial fish species.

This regime is characterized by a substantially reduced abundance of the valuable fish species. In cases where the valuable fish species performs key ecological functions, it can lead to trophic cascades that lead to an overabundance of planktivorous fish or primary producers such as seagrass, macroalgae, sponge or phytoplankton (Jackson et al. 2001). With low fish abundance, food production is the most immediate ecosystem services affected. While industrial fishermen often can move and harvest new areas or new species, artisanal fishermen often can lose their livelihoods and traditions.  

There are two key direct drivers of fisheries collapse: overfishing as a top-down disturbance, and the change in nutrients input as a bottom-up disturbance.

Overfishing reduces the population size of commercial fish species, reduces their ability to perform their functional role in the ecosystem, can destabilize their population dynamics (Anderson et al. 2008), and ultimately could change the structure of the food web. Nutrient inputs, on the other hand, affect fisheries by increasing the abundance of lower trophic levels e.g. phytoplankton and zooplanktivores such as jellyfish (Daskalov et al. 2007). Such changes in food web configuration can reduce the energy transfer to fish and hence reduce fishing productivity. These drivers can reduce the resilience of a regime making it more vulnerable to other types of regime shift, such as climate driven food web reorganization [see: Climate & Marine Food Webs]. The indirect drivers that explain overfishing and changes in nutrient input interact in synergistic ways, making their identification a challenging process (Ocean Studies Board 2006, Kirby et al. 2009).

Indirect drivers of overfishing include oversupply of fishing boats, inflexible fishing quotas, demand for fish, the development of new fishing technologies, the facilitation of trade (market connectivity), governance failures such as the failure of regulation to halt illegal fishing, and market failures such as subsidies or perverse incentives. Nutrient enrichment of marine ecosystems is also indirectly driven by runoff of fertilizer nutrients from agriculture and sewage from urban areas (Diaz and Rosenberg 2008).

A fishery collapses when a high-valued commercial species is substantially depleted and fails to immediately recover when fishing effort is removed (Hutchings 2000). It is usually associated with a substantial and persistent change in the structure of the marine community. Such a change involves changes in the species presence, their relative abundance, size and age.

When the stock of commercial species decreases or even disappears the populations of other similar species may increase. Many commercially fished species are predatory fish, such as tuna, that have a high trophic level. Removing such top predators can cause a trophic cascade to reshuffle a food web (Daskalov et al. 2007, Estes et al. 2011), especially if there are not ecologically similar species present. Such trophic cascades, can increase fish stocks of prey species (Walters and Kitchell 2001), which can in turn lead to either sequential depletion of fish stocks or the development of new fisheries (Essington et al. 2006).

The low abundance of commercial fish species can be maintained by a number of mechanisms. First, when the population of the commercial fish species falls under certain levels their population growth rate can decline from a variety of causes. This is known as depensation in fisheries, or the Allee effect more generally (Liermann and Hilborn 1997). Such an effect can occur if small schools of fish offer poorer protection from predators, smaller populations are less able to hunt or detect food, or mating success rates can decline.

Second, nutrient enrichment can favor the food web shifts towards zooplanktivores, such as jellyfish. By reducing resources for other fish, the rise of such species can suppress fisheries recovery. Fishery pressure on filter-feeding fishes like sardine exacerbates the problem by favoring jellyfish outbreaks (Jackson et al. 2001, Brotz et al. 2012, Condon et al. 2013).

Despite evidence from some well-documented cases, the degree to which fisheries declines are regime shifts remains a controversial issue for a number of reasons (Hilborn 2007, Litzow and Urban 2009). First, time series for most fisheries are only available from 1950, a time when marine ecosystems were already affected by flourishing industrial fishing, and some stocks had already been seriously depleted, especially those of large marine vertebrates like whales, manatees, crocodiles, seals, swordfish, sharks and rays (Kirby et al. 2009). This lack of long term data makes it difficult to identify causation in fisheries time series. Furthermore, marine ecological diversity and food web complexity complicates the detection of changes in fish populations due to fishing, because the signals of regime shift dynamics are small relative to ecological variation (Liermann and Hilborn 1997).

Shift from high to low abundance of commercial fish species

Since 1950 the world marine fisheries catch has expanded almost fivefold, and is currently about 80 million tonnes of fish. According to FAO, about 30% of fish stocks are overexploited, 60% are fully exploited and about 10 % are not fully exploited (FAO 2012). Global marine fish catches have plateaued and are not expected to increase. Local fisheries decline has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Pauly et al. 2006). There is not clear assessment of the number of stocks that have collapsed but FAO suggests about 5% of fish stocks are depleted (FAO 2012). Some collapsed stocks have remained at under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the collapse of fisheries.  

Shift from low to high abundance of commercial fish species

An abundant fish population regime will likely better maintain marine biodiversity and will be more resilience to climate variation and climate change. Society will benefit of high abundance regime in terms of high primary productivity that translate in more food, jobs and livelihoods in small coastal communities. It will also better regulate marine hypoxia.

A review of marine fisheries in the Millennium Ecosystem Assessment offer a synthetic set of managerial options for addressing fisheries collapse (Pauly et al. 2006).

Options for enhancing resilience

There are two types of strategies for enhancing resilience. The first is to stop doing things that are eroding resilience, while the second is to being activities that build resilience. There are a number of fishing practices that could be eliminated because they damage the resilience of fisheries. For example, fishing practices that destroy habitat such as bottom trawling should be avoided. Developing alternative technologies and reduction of fishing effort are suggested. Bycatch, the killing but not consuming of unwanted fish, can also have substantial impacts and can be reduced. Indirectly, fisheries subsidies and inflexible quotas can drive fishing effort to high levels that erode the resilience of a fishery.

Options for reducing resilience of unwanted regime to encourage restoration or transformation

Direct resilience building strategies include the establishment of Marine protected areas (MPAs) (Takashina and Mougi 2014). These areas can play an important role by providing refugia for fish and increasing ecological diversity. Research suggests that they need to cover ecologically significant areas, including different habitats, ensuring connectivity among them. However there remains controversy over the extent to which MPAs shift fishing pressure to less protected sites.

Indirect resilience building efforts include efficient regulation and policing of fishing can reduce illegal fishing, while comprehensive fisheries policies at the international and national level can also build resilience if they result in fisheries policies that reinforce one another. Indirect resilience building strategies include improving the ability of fishers to exit fisheries.

Good management of fishing pressure has been able to restore fisheries stocks. For example, in the US fishing closures and implementation of management measures for bottom fishing have enabled the restoration of George Bank haddock, Atlantic scallops, George Bank yellowtail flounder, Atlantic striped bass, Atlantic Arcadian redfish, Pacific chub mackerel, and Pacific sardine. However, there is no clear strategy that works across many ecosystems or nations (Beddington et al. 2007).

Mangroves are ecosystems adapted to the mixture of fresh and salt water common in intertidal zones of marine coastal environments. Mangroves cover between 1.7-1.8 x 105 km² globally. They develop within 30ºN and 38ºS and grow optimally between 15ºC and 25ºC (Lovelock 2008). Outside this geographical and thermal range mangroves show reduced leaf formation and above 38ºC or below freezing temperatures they do not photosynthesize (Mcleod and Salm 2006; Cavanaugh et al. 2014). Mangrove ecosystems are often used by local communities who depend on fishing and timber harvesting.

Mangrove forest

Mangrove forest is characterized by an ensemble of species with aerial roots adapted to survive in environments where water salinity and tides are highly variable. Mangroves grow in tropical and subtropical waters, where the optimal temperature range is between 15-24ºC. Different species occur under different salinity gradients and some trees reach 40m high in areas with high precipitation. Mangroves usually establish in soil enriched by the sediment discharge of rivers (Mcleod and Salm 2006). Mangroves are ecologically important because they provide habitat for many marine species at some stages of their lifespan, increasing both marine and terrestrial biodiversity.

Salt marshes, rocky tidal, shrimp farms.

Mangroves can become dominated by salt marshes or other configuration of coastal wetlands depending on the substrate, sediment deposition and the topography (Mcleod and Salm 2006). The alternative regime once a mangrove has collapsed is not straightforward. It strongly depends on the land use, topological characteristics, latitude and the history of the system disturbance (Cavanaugh et al. 2014). For example, many mangroves around the world have been converted into ponds for shrimp farming, salt extraction, or simply dry out to establish agriculture, cattle ranching, infrastructure development or human settlements.

The main drivers of mangrove collapse have typically been land use/cover change during the last 50 years. One third of world’s mangroves have been lost due to overexploitation of forest resources (deforestation) or conversion into agriculture, salt extraction, or ponds for shrimp farming (Cavanaugh et al. 2014). Other threats to mangroves include the development of infrastructure (roads, dams), the diversion of fresh water for irrigation and the development of urban areas. These impacts are typically stronger in developing countries where it is expected that mangroves will decline 25% by 2025 (Mcleod and Salm 2006), while recent studies show that mangrove areas are declining 1-2% yearly (Duke et al. 2007; Alongi 200; Cavanaugh et al. 2014).

Current climate change is also expected to affect mangroves via the increase in frequency and severity of storms, floods and droughts (climate extreme events), the increase of temperature, sea level rise, and ocean acidification. Frequent extreme events that potentially decrease the flow of fresh water will increase salinity, putting stress on some species and reducing their growing rates and seedling survival (Mcleod and Salm 2006). Precipitation increase, on the other hand, could favor mangrove over salt marshes by decreasing salinity and giving a competitive advantage to mangroves (Cavanaugh et al. 2014). However, strong storms and flooding events could prevent mangroves from respiring when their aerial roots are underwater. Although temperature is expected to increase 2-6ºC before 2100, its effect on mangroves is contested. On the one hand, a 6ºC increase would heterogeneously affect mangroves around the world, although it is likely to induce thermal stress, it won’t be strong enough to surpass the temperature tolerance of these ecosystems. An increase in temperature will increase the melting rate of ice and expand the volume of world’s oceans increasing the level of the sea. On the other hand, it has also been observed that warming on the coldest time of the year can favor mangroves over taking salt marshes in temperate areas of the planet, with a suggested threshold of -4ºC for Florida (Cavanaugh et al. 2014).

Sea level rise is perhaps the most important threat to mangroves. While the expected sea level rise projection range from 0.09 to 0.88 m (or 0.9 to 8.8 mm per year) several studies report that mangrove ability to migrate upwards range from 1.2 to 4.5mm per year (Mcleod and Salm 2006). Although faster migration has been observed in Holocene stratigraphic records for Florida, migration also depends of local topological conditions. Increase in atmospheric carbon dioxide could increase mangroves growth but also ocean acidification. While the increase in growth is not expected to be strong enough to overcome the pressure of sea level rise, acidification is expected to have an indirect effect on mangroves by reducing coral reefs accretion, increasing in turn the erosive effects of waves on mangrove soils (Mcleod and Salm 2006).

Shift from mangroves to salt marshes, rocky tidal or shrimp ponds.

By having aerial roots mangroves adapt to very specific conditions where salt and fresh water meets in coastal zones. Mangroves build up carbon rich soils also known as peat, which favors further mangrove growth.

Loss of mangrove area decreases peat production. Peat is lost by wave erosion and reduces the likelihood of recovery of mangrove forest. Mangrove area loss is mainly driven by conversion to agriculture, urban settlements, overexploitation of wood, shrimp farming, salt extraction, or infrastructure development (e.g. roads, dams, water divergence to irrigation) (Cavanaugh et al. 2014). Human perturbations are maintained by social feedbacks such as demand for food and market access, population growth and the cumulative benefit of human settlements (higher provision of services). In contrast, communities with poor services access such as electric energy or gas for cooking strongly rely on mangrove wood as source of energy and construction material (Mcleod and Salm 2006).

If sea level rise occurs faster than the ability of mangroves to build up peat or migrate to higher areas, mangrove forests will disappear in major areas of the world (Alongi 2008). Both the impact of sea level rise and mangrove’s ability to adapt depend on local conditions and are highly uncertain. For example, mangroves in carbonate substrates (such as coralline islands) and microtidal systems (short tidal change from high to low) are less likely to adapt and migrate as climate change continues. Mangroves in rich soil substrates and macrotidal systems are more likely to migrate if salinity balance and sedimentation do not overcome their tolerance range (Mcleod and Salm 2006).  

Shift from salt marshes to mangroves  

Mangroves have also been observed to migrate pole wards colonizing habitats previously dominated by salt marshes (Cavanaugh et al. 2014). The shift occurs when the minimum temperature in winter increase for a longer period of time, giving mangroves an advantage to over compete salt marshes. It has been observed in areas such as Florida, Louisiana or Australia. Although there is uncertainty regarding the mechanism, the threshold is thought to be related with the number of days that minimum temperatures are colder than -4ºC, at least for Florida case (Cavanaugh et al. 2014). The authors of the Florida study also report that it’s unlikely that local scale drivers such as nutrients inputs, sedimentation or hydrology have had a determinant effect on the shift. However, sea level rise has contributed to increasing mangrove area (since it has happen slowly enough) and other common driver of the shift can reduce mangrove accretion such as coastal development, aquaculture and timber production (Cavanaugh et al. 2014).

Mangroves provide a variety of ecosystem services. Mangrove ecosystems provide habitat for economically important marine species for fishing (e.g. shrimp or lobster among other crustaceans and mollusc), and they provide habitats for important groups of terrestrial species such as reptiles, monkeys, and birds. They are therefore important for maintaining both marine and terrestrial biodiversity. Mangroves also provide fuelwood, charcoal, and wood for construction. As a forest it provides protection against storms, floods, river born siltation and also traps water pollutants and sediments, reducing the erosive action of waves, particularly protecting adjacent ecosystems such as coral reefs, and sea grass beds (Duke et al. 2007). Mangroves have also been shown to play a key role in the global carbon budget, capturing up to 15% of global carbon and exporting up to 11% of particular terrestrial carbon to the oceans (Alongi 2014). In fact, recent studies support the hypothesis that mangroves are able to capture and store carbon in a greater extend that terrestrial ecosystems (Lovelock 2008), nearly identical to those of tropical forest (Alongi 2014). Therefore, losing mangrove area implies the loss of carbon storage.

While in the short term local communities could benefit by converting mangroves into shrimp farms, agriculture fields, or using its wood; in the long term they mean the loss of important services such as coastline protection, biodiversity important for tourism and fishing, water cleansing and carbon storage (Ewel et al. 1998; Cavanaugh et al. 2014). In fact, their services has been valued over $1.6 trillion per year (Costanza et al. 1997).

McLeod et al. (2006) propose a series of tools to monitor and plan for mangroves adaptation to climate change. First they propose to assess mangrove vulnerability based on the local conditions. Management options for mangrove forests are highly context dependent. As the main threat to mangroves is climate change through sea level rise, mangrove ecosystems will likely migrate upwards and northwards (Cavanaugh et al. 2014). However, such migration is constrained by local conditions such as substrate types, infrastructure development, sediment input and changes in salinity.

Second, they propose to apply risk-spreading principles (Duke et al. 2007) and identify areas that are likely to be sources and sinks of the migratory process (Mcleod and Salm 2006).  By identifying the role of each mangrove patch, it is easier to prioritize where green belts are needed to increase connectivity, which patches should be protected and which patches are more likely to respond to restoration efforts. They emphasize the importance of establishing a baseline and monitoring program with local communities in order to assess the main drivers locally. Developing partnerships with local stakeholders and developing alternative livelihoods for mangrove dependent communities are key managerial efforts (Mcleod and Salm 2006). The resilience of mangroves to future climate change scenarios also depend on how successful management efforts are at reducing the influence of other drivers: deforestation, overexploitation, infrastructure development, divergence of water for irrigation, and urbanization. Management of such local forcing will increase resilience to regional forcing such as increasing storm events or ocean acidification.

Third, it has been suggested that policy and market-based are critical to foster conservation efforts, reduce mangrove encroachment and maintain carbon storage through funding mechanisms such as REDD or other payment for ecosystem services schemes (Hutchison et al. 2013). Effective governance structures and education has also been suggested as managerial strategies (Duke et al. 2007).

Co-operation and sustainable resource levels

Harvesters extract from the resource at rates that are socially optimal, ensuring that resource levels stay at their most productive level and the community-average payoff is high. Social capital exists within the community, ensuring any defectors are ostracised.

Over-harvesting

Harvesters exert high efforts in extracting the resource, leading to depletion of the resource and low payoffs for the community. Social capital is absent and defectors are not ostracised.

Shift from ‘Co-operation and sustainable resource levels’ to ‘Over-harvesting’

The regime shift occurs when ostracism ceases to be an effective mechanism for encouraging defectors to co-operate, because the benefits of overharvesting begin to outweigh the disadvantages of ostracism. A number of factors could drive this shift. Increasing resource level, for example due to increased inflow, can lead to ineffective ostracism because at higher resource levels defection becomes more attractive by providing higher gains from resource over-extraction. Increasing defector payoff compared to ostracism, for example due to decreased costs or decreased ostracism strength, could also lead to defection becoming increasingly attractive.

Shift from ‘Over-harvesting’ to ‘Co-operation and sustainable resource levels’

Reversal of any of the above trends can cause a shift from over-harvesting to co-operation: decrease in resource levels, for example due to decreased inflow; increased costs; or increased ostracism strength.

Co-operation and sustainable resource levels occur when there is sufficient social capital to ostracise defectors and thereby make defection less attractive than co-operation. The key negative feedback that keeps the regime stable is the following: if the fraction of cooperators increases the social capital and hence strength of ostracism increases which reduces the utility of defectors, making defection less attractive. At the same time, however, resource productivity increases which increases the utility of defectors, so stability ultimately depends on the respective strength of the two feedback loops.

The over-harvesting regime exists when there are only few cooperators and hence there is insufficient social capital to ostracise defectors. The feedback that maintains full defection is as follows: any agent that 'co-operates' (i.e. harvests at a lower level) will have a payoff substantially less than the defectors. In the absence of social capital to encourage co-operation through ostracism, the co-operator will switch back to defection. The key threshold for a collapse in co-operation is when the costs of ostracism are lower than the benefits of overharvesting, i.e. ostracism ceases to be an effective mechanism to discourage defection. A variety of drivers can cause this shift, as discussed below.

To trigger a transition from full defection back to co-operation, co-operation must become more attractive than defection. This could be achieved by increasing the costs of harvesting, or by decreasing the level of the resource to low levels so that the benefits of overharvesting are minimal. However, once the co-operation strategy is lost in the community, it may be very difficult for it to re-emerge.

The ability of the natural system to provide resources for harvesting is lost with the regime shift from co-operation to over-harvesting (gained for over-harvesting to co-operation). Depending on context, loss of other ecosystem services many accompany the decline in resources. The income that members of the harvesting community obtain by harvesting is severely decreased by this regime shift. Income may even no longer exceed the costs of harvesting and the community may need to find other means of survival.

Options for preventing regime shift to over-harvesting

Management actions that stop the drivers discussed above reaching their thresholds may help to prevent regime shifts. Activities that strengthen social norms and trust in the community and thus enhance cooperation and decrease the incentive to defect and overharvest for the individual benefit (hence increasing the strength of the ostracism).

Options for restoration of co-operation:

Introducing measures that build trust in the community and disincentivise free riding. 

Steppe

The steppe vegetation coverage is dominated by grasses growing in dry soils (Eroglu et al. 2012). In this regime, herbivores increase nutrient availability by accelerating nutrient cycling. This increases primary production. During the Pleistocene epoch, arctic steppes sustained large grazing herbivores such as mammoths, bison and yaks (Zimov 2005). Today, grasses still sustain reindeer/caribou population. Currently, the steppe communities of Yakutia, Siberia are confined to the terraces above floodplains and south facing slopes of the river valleys (Yurtzev 1982).

Tundra

The tundra vegetation coverage is dominated by mosses and shrub such as dwarf birches and willows growing in waterlogged soils. The ecosystem has lower primary production and nutrient cycling thus the growth rate is slow. Moss growth is limited by moisture so the moist and nutrient low soil promote and sustain moss growth.

Herbivory has caused large-scale vegetation changes across the Arctic (Speed et al. 2009; Bråthen et al. 2007; van der Wal 2005; Srivastava & Jefferies 1996). Herbivores trample the vegetation and deposit faeces to nutrient limited soils which promotes grass growth sustaining herbivore population. In the absence of herbivores grass litter increases causing soil moisture to increase which reduces nutrient availability promoting the growth of unpalatable mosses.

Climate can act as a driver by changing air temperatures which is associated with changes in soil moisture. Soil moisture is key component in sustaining moss or grass coverage. Moss growth is more limited by water than by nutrients. However, nutrients are more available in drier soils which promote grass growth. Currently, climate is changing, in the long term whether steppe, tundra or another regime will dominate Arctic ecosystems in uncertain.

Shift from Steppe to Tundra
Grass-dominated steppe is maintained by trampling and nutrient inputs from herbivore faeces. The enriched soils promote the growth of grass coverage (Shaver et al. 1986). Grasses have a strong influence on soil moisture as they dry out the soil through high transpiration (Zimov 2005). Dry soil conditions are associated with increased nutrient availability intensifying grass growth (Nadelhoffer et al. 1991).

In the absence of herbivores, moss coverage increases creating water-logged soils (Zimov 1995). Water-logged soils prevent decomposition of organic matter, so nutrient availability decreases, which limits the growth of grasses and, hence, promotes moss growth instead (Zimov 1995). As moss growth increases, herbivore density is further reduced because moss have lower nutritional value.

Moss-dominated tundra is susceptible to trampling, a disturbance caused by large herbivores; therefore, moss coverage increases with low herbivore density (van der Wal 2001). Mosses causes an increase in soil moisture. The combination of high soil moisture creates the conditions for water-logged soils establishing optimal conditions for moss growth (Zimov 1995).

Shift from Tundra to Steppe

Climate is the main driver that could shift tundra back to steppe; this is because large grazers don't exist any more and arctic ecosystems are vulnerable to climate change. Rising of air temperatures would deepen the thawing layer of permafrost and increase evaporation. Therefore, frozen nutrients from soil would become available, which would promote grass growth; in addition to that, higher temperatures will enhance organic matter decomposition and hence nutrient availability. Then, grasses will increase evaporation. Nevertheless, moist soil conditions might be resilient since permafrost melting will make frozen water available. If there are no draining conditions, moisture will provide a suitable environment for mosses that will insulate soil temperatures and then buffer permafrost melting.

Rising of temperatures will create dynamics on arctic ecosystems that are difficult to predict. However, recent evidence shows that tundra could turn into different types of vegetation (e.g. shrub land, forest, lake, grassland) depending on local conditions (Karlsson et al. 2011). Steppe-grasses will prevail on drier areas, but more probably as patches within a mosaic since large herbivores that maintained steppe conditions during the Pleistocene, are not present.

Shift from Steppe to tundra

Shifting from steppe to tundra had a great impact on provision of food. The thermal insulation of the tundra prevents permafrost melting, which prevents destabilization and collapse of infrastructure (Schaefer 2012). Increasing air temperature due to climate change deepens the active layer of permafrost. However, soils covered by moss, permafrost is less susceptible to degradation (Ivanova 2003), this ecosystem service is essential since permafrost contains almost twice as much carbon as the atmosphere today and its effects would be irreversible at human time scale (Schaefer et al. 2012).

The wellbeing of nomadic indigenous peoples such as Nenets, Enets, Sami, Nganasans and Selkups, are connected to Arctic that provide opportunities for grazing and food sources, as well as ecosystems that co-produce cultural ecosystem services i.e. production and conservation of indigenous knowledge, practices and believes, as well as production of scientific knowledge (Huntington 2013). Tourism is also a cultural ecosystem service that will be affected.

Tundra to steppe

Arctic soils covered by mosses, permafrost is less susceptible to degradation (Ivanova 2003). Transition from tundra to steppe implies loss of this buffering layer. Consequence, the following ecosystem services will be impacted: Water availability. Local and regional freshwater dynamics will change leading to a decline in the number of lakes and wetlands (Wrona et al. 2016; Schaefer 2012). Increased abundance of grassland will affect livestock production, steppe might benefit both reindeer hunters and herders. Protection from infrastructure destabilization; pipelines, railways and power lines across the arctic are built on solid ground provided by the permafrost that if degraded their stability might be lost. This might cause economic costs and serious ecological damages (Schaefer, 2012). Fire regulation; as moss tundra shifts to a drier regime, possibilities of wild fires will increase, which contributes to thermal erosion of permafrost (Schaefer, 2012). Prevention of climate change; permafrost degradation turns permafrost's carbon stock into a carbon source, which would accelerate climate change (Schuur et al. 2015). Loss of this ecosystem service impacts the global level, since permafrost contains twice as much carbon as the atmosphere and effects of release is irreversible at human time scale (Schaefer, 2012).

Cultural ecosystem services will also be affected; conservation of indigenous knowledge, practices and believes related to tundra would be lost, affecting the well being of t he Nenets, Enets, Sami, Nganasans and Selkups (Huntington 2013).

To prevent the steppe from shifting into tundra, greenhouse gas emissions must be reduced in order to avoid the effects of climate change. The planetary boundary value for atmospheric CO2 concentration is 350 ppm above pre-industrial level (Rockstrom et al. 2009). This was further supported by the COP21 agreement, a treaty signed to limit temperature increase to 1.5^oC compared to pre-industrial levels. These efforts are not enough to prevent the feedback loop, eliminating emission is the only solution. In the Artic, land management strategies can address feedback loop on the local level. In a study, by Post and Pederson (2008), a 5-year experimental investigation of Arctic plant community response to warming, showed that warming under continuous grazing pressure from muskoxen did not differ from plant composition without warming suggest that management of large herbivores may be an important aspect for mitigating ecosystem response to future climate change

Attempts are being made today to restore the Pleistocene steppe ecosystem on a limited spatial extent in Siberia. Sergey Zimov's Pleistonce Park, 160 km², attempts to reestablish the steppe ecosystem by reintroducing megafauna to Northern Siberia (Republic of Yakutia). Zimov (2005) and colleagues tried to test if the 'key-herbivore´ hypothesis can be verified, as today's Holocene climate should be optimal for the Pleistocene steppe vegetation. The megafauna consisting of reindeer, moose, Yakutian horses, musk oxen and bison would influence the vegetation and soil composition, by trampling on grassland and returning nutrients to the soil through their manure (Zimov 2005). The grass root systems stabilize the soil and trampling reduces the albedo, exposing ground to colder temperatures, both would prevent permafrost from melting.

Due to thermokarst processes such as permafrost melting, numerous areas of the Arctic landscape are extremely lake-rich.  The water balance of those lakes may either be maintained due to impermeable permafrost that limits groundwater flow (Marsh 2009) or lost due to seasonal drainage, a natural characteristic of the Arctic system. However, a widespread, unnatural decline in lake abundance is being observed within the Arctic due to decades of rising atmospheric temperatures in the region that has been deteriorating the state of the permafrost (Smith et al. 2005). The lake-rich landscape of the Arctic is therefore slowly transforming into a drained, dry landscape increasingly dominated by vegetation such as shrubs and graminoids (Hinzman 2005).

Thermokarst lake ecosystem

The thermokarst lake-dominated landscape is formed as a response to some imposed disturbance, such as road construction, wildfire or climatic warming, which causes permafrost to thaw, creating irregular surface topography depressions, called thermokarsts. Those may appear as near-surface massive ice melts, allowing the surface to subside (Hinzman et al. 2005). As permafrost melts over time, those depressions become filled with melted water forming the lakes characteristic of the Arctic region.

Terrestrial ecosystem

Driven by external and internal factors, Arctic lakes are drying up at an unnatural speed within the region. This shift is accompanied by changes in vegetation, for example, increased shrub- and graminord abundance and a decrease in the cover of mosses and lichens through a broad expanse of the Arctic (Hinzman et al. 2005). The changes in vegetation may then further transform the landscape through sequential effects on foraging mammals and birds, as well as aquatic fauna and insects (Hinzman et al 2005).

Shift from thermokarst lakes to terrestrial ecosystem

The main external driver of this regime shift is climate change, which is caused by elevated greenhouse gas emissions (IPCC 2007). Increased temperatures warm the soil and warms the permafrost. As the permafrost thaws, greenhouse gasses such as CO² and CH₄ that are trapped in the ice are released into the atmosphere, further increasing the effects of climate change (Hinzman 2005; Karlsson et al. 2011). At a local scale, the effects of climate change are incremental. However, as this interaction is occurs over large areas, the effects are aggregated and self-reinforcing, which is significant since permafrost and seasonally frozen grounds store about 25% of the total global soil carbon stock (IPCC 2007).

At a local scale, initial thawing of the permafrost from increased atmospheric temperatures creates a dynamic landscape of depressions (thermokarst) in which lakes are formed (IPCC 2007). As permafrost become more degraded, drainage through outlet channels occurs. Since the temperature of melted water is higher than that of ice, it will exacerbate the melting of permafrost, causing another reinforcing feedback. Once the layer of frozen ground is completely penetrated the lakes can become permanently drained allowing for vegetation to be established (Marsh et al. 2008).

Shift from terrestrial ecosystem to thermokarst lake ecosystem

Heightened air- and soil temperatures melt the ice-rich soils and permafrost on which forests in the Arctic region develop, changing the physical conditions for the forest growing on top of it. When the roots of the trees get flooded, the trees die and ponds and lakes eventually replace the forest (Hinzman 2005). However, it is important to note that the shift from a terrestrial ecosystem to a lake dominated one is not as frequent as the shift from lakes to terrestrial ecosystems. The reason for this are the reinforcing feedbacks as mentioned above.

The warming of soil temperatures leads to the melting of continuous permafrost and the subsequent increase in abundance of discontinuous or sporadic permafrost. Melting of permafrost causes depressions in the soil called thermokarsts. Increased precipitation in the form of snowfall may have either an insulating or cooling effect on soil temperatures depending on several variables such a timing, duration, accumulation, and melting processes of seasonal snow cover, density, structure, and thickness of seasonal snow cover etc. (Zhang 2005). Depending upon the local surface energy balance, the thawed ground may refreeze or the permafrost can continue to degrade (Hinzman et al. 2005). On sites where continuous permafrost is thawing the active layer of soil in which melted water can move deepens allowing for the formation of thermokarst lakes (Hinzman et al. 2005). In freezing seasons, if the temperature is low enough and the hydraulic parameters on site suitable, the formation of these lakes can contribute to the re-formation of permafrost soils. Ice-rich permafrost can also prevent surface water from infiltrating to deeper groundwater zones, causing surface soils to be very wet. Decreased albedo due to changes in vegetation, the extension of snow-free and ice-free periods on terrestrial and lake surfaces and abundance, and reduction in the area occupied by glaciers and continental ice sheets in high latitudes may act as a positive feedback into atmospheric warming (Hinzman et al. 2005).

On sites where there is an abundance of discontinuous permafrost the active layer also deepens allowing for increased surface area for sub-ground water movement that increases the rate of drainage. Thermokarst lakes and ponds may begin to fill or drain depending upon the direction of the hydraulic gradient beneath the lake (Hinzman et al. 2005). Therefore it is not the depth of the active layer alone that determines the rate of drainage. However, increased drainage of lakes may prevent the reformation of permafrost from lack of available water for freezing in winter months and it can be inferred through advection of heat from decreased albedo. Once these drained lake beds become permanently drained they are likely to be re-vegetated which will change the albedo and may cause the extension of snow-free and ice-free periods and may act as feedback into increasing atmospheric temperature (Hinzman et al. 2005).

The affects of the shifts from lake to terrestrial ecosystems are aggregated at a local, regional and global scale due to the large scale and frequency at which those dynamics occur. The whole regime as described in the causal loop diagram below is a re-enforcing feedback that at its current state is leading to an overall decrease in lake dominated ecosystems and a increase in terrestrial ecosystems (Smith et al. 2005).

Even though this regime shift occurs at the local scale, its consequences go beyond that scale because it reshapes regional hydrology (through changes in water balance and surface water- connection and fragmentation) and alters regional supplies of the ecosystem (Karlsson et al. 2011). This regime shift therefore affects provisioning ecosystem services, as it will decrease the availability of freshwater. The above-ground stored freshwater, which is of vital importance to migratory birds, fish, and other wildlife used by indigenous people, will shift to below-ground stored freshwater (Artic Science Journeys Radio Stories 2005). Globally, however, freshwater will become a more abundant ES due to reductions permafrost and subsequently ice sheets and glaciers resulting from the regime’s feedbacks to climate change. Nevertheless, this global increase in the freshwater ES is not a positive one: “Increased freshwater delivery to the Arctic Ocean from reductions in ice sheets and glaciers result in rising sea level (...)” (Hinzman et al. 2005). At a local scale traditional hunting and fishing practices will be impaired (Vincent et al. 2013).  Arctic freshwater systems provide important migratory routes for fish stock. Due to this regime shit, those routes would be greatly altered in connectivity among lakes and river channels, as well as in terms of their physical coupling to the coastal marine ecosystem (Vincent et al. 2012). Globally, fishing practices would suffer as well. Increased discharge of water into the Arctic Ocean due to draining results in increased nutrient and external organic matter inputs to the Arctic Ocean that affect primary processes at the base of the marine food web (Vincent et al. 2012). Climate regulation is heavily affected on a global scale through the processes of this regime shift: as it sets up a positive feedback cycle in which the release of methane and carbon dioxide through the newly exposed soil and melted permafrost feeds into global warming. The shift from lakes to terrestrial ecosystems heightens gully erosion, a type of erosion that occurs when water is channeled across unprotected land and washes away the soil along the drainage lines. As permafrost degrades further, it causes erosion and slumping of lake edges and stream channels

The Arctic lakes are of vital importance to indigenous people as they are a source of fresh drinking water and also provides important winter transport routes. These lakes not only allows indigenous people access to their traditional hunting and fishing areas, but also to transport goods to remote communities and industries such as mining centers (Vincent et al. 2013). The regime shift would also mean an increase in products extracted from woods such as timber commodities and fuel: “increased areas of tree growth in the Arctic could serve to (…) take supply more wood products and related employment, providing local and global economic benefits” (Hassol et al. 2004).

There are few management options available to maintain the thermokarst lake ecosystems due to the long time scale at which permafrost melts and the natural dynamics within this system e.g. hydraulic fluctuations or site-specific soil conditions. One management option is to continue to reduce the emissions of greenhouse gases while simultaneously investigating further opportunities for carbon offsetting schemes.

For hydroelectric reservoirs, shifting ice conditions will have both positive and negative effects, and may require adaptive changes in operating procedures, with attention to minimize negative impacts associated with ice jams and ice breakup downstream of the spillway (Vincent et al. 2013). Fisheries management plans will also need to be adapted to the changes in migration and productivity of northern fish populations with ongoing climate change (Vincent et al. 2013). These essential resources require the development of integrated freshwater management plans, which include consideration of alternate water sources as traditional supplies change in quantity or quality (Vincent et al. 2011). The Arctic will require increasing vigilance and appropriate water management strategies to avoid and minimize the impacts of changing water impacts in the future (Vincent et al. 2013).

The boreal forest in northern Interior Alaska covers an approximate area of 48 million hectares (Mann et al. 2012). Bounded by the Alaska Range to the south, the Seward Peninsula to the west and Brooks Range to the north (Wolken et al. 2011). It is characterized by large areas of gently sloping uplands and flat lowlands isolated by mountain ranges and braided rivers with broad floodplains. During the past 50 years, the mean annual temperature in the interior boreal forest of Alaska has increased by 1.3 ºC (Hartmann and Wendler, 2005) and this has been strongly associated with recent increase in fire frequency and severity (Wolken et al. 2011). The annual area burned in Interior Alaska has doubled during the last decade (Kasischke et al. 2010). The forest has been dominated by coniferous trees (mainly black spruce), which is a group of species that thrive under colder, moist conditions in a deep soil organic layer. However, due to climate warming, Interior Alaska is currently shifting towards a deciduous dominated forest (e.g. aspen and birch) that prefer warmer and drier conditions with a shallow soil organic layer (Johnstone et al. 2010).  

Low fire frequency coniferous dominated forest

Under cold and wet conditions, coniferous trees, such as the black spruce (Picea mariana), dominate the boreal forest in Interior Alaska. The coniferous trees promote the accumulation of a deep soil organic layer, cold and moist soil temperatures, and slow decomposition rates. Due to the moistness of the soil, fires are not frequent in these forests. However, when they occur, they tend to be severe due to the high flammability of the black spruce. Due to the high water content of the soil organic layer, it will only be entirely consumed by the most severe fires (Johnstone et al. 2010). As long as the entire organic layer has not been consumed, the coniferous trees have a regenerative advantage due to their seed’s ability to sprout in moist organic soils. This means that these forests tend to return to a coniferous dominated state even after a fire (Hollingsworth et al. 2013).

High fire frequency deciduous dominated forest

A deciduous dominated forest, mainly consisting of aspen and birch trees in Interior Alaska, prefers and reinforces dry and warm soil conditions. Although these forests have a high organic litter production, the fast decomposition rate and the subsequent fast nutrient turnover, leaves a shallow soil organic layer and a nutrient rich mineral soil that favours deciduous tree regeneration. The warm and dry conditions in the forest lead to a comparatively high fire frequency, but due to the low flammability of the deciduous trees, the fires rarely become very severe (Johnstone et al. 2010). However, because of the shallow pre-fire soil organic layer, the fires tend to expose the mineral soil, which favours the post-fire reproduction of deciduous trees (Hollingsworth et al. 2013).

The main external direct driver of the regime shift is climate warming, which affects the important feedbacks of the boreal forest ecosystem. Warmer and drier climatic conditions have a direct effect on the most important soil characteristics (i.e. soil temperature, moisture and depth of the soil organic layer), influencing the vegetation cover. Coniferous trees require a cold moist soil with a deep soil organic layer to successfully reproduce. Deciduous trees, on the other hand, are competitively superior under warmer and drier soil conditions with a shallow soil organic layer (Johnstone et al. 2010). Extended summer seasons and the accompanying drier conditions create a suitable environment for an increase in the annual number of natural ignitions by lightning strikes (Kasischke et al. in press). Fires have an immediate effect on the environmental conditions of the forest, affecting factors such as soil moisture and the post-fire forest community composition (Kelly et al. 2012). Once established, the trees support and emphasize the conditions that favour their own reproduction and dominance. Under colder environmental conditions, coniferous trees, such as the black spruce, produce large amounts of organic litter, which, together with the presence of permafrost, maintains a moist soil and slow decomposition rates, contributing to a deep soil organic layer. Under warmer climatic conditions, deciduous trees produce ground layer fuels that decompose at fast rates, maintaining a shallow organic layer (Johnstone et al. 2010). Global socio-economic development indirectly drives climate warming, contributing to the increased atmospheric temperatures through e.g. greenhouse gas emissions (IPCC, 2011). Fire management can have a limited effect on the regime shift through fire suppression practices.

Shift from coniferous to deciduous-dominated forest

The coniferous dominated boreal forest of Interior Alaska requires a deep soil organic layer to succeed over other classes of trees. Once the required depth is present, this has a strong reinforcing effect on the coniferous forest. This is a requisite to maintain cold moist soil conditions in which coniferous trees thrive. The cold soil temperature reduces the decomposition rate of organic litter, maintaining a deep soil organic layer (Johnstone et al. 2010).

External shocks such as severe and frequent fires or insect outbreaks might cause dramatic shifts in the soil and vegetation by impacting the soil organic layer depth and tree succession (Hollingsworth et al. 2013; McCullough et al. 1998; Wolken et al. 2011). If the severity of such external shocks is sufficient, thresholds in the most important regime characteristics will be crossed (e.g. shallow post-shock soil organic layer) creating windows of opportunity for deciduous species to dominate. Atmospheric temperature influences the frequency and severity of fires, the soil characteristics and the length of the growing season; therefore, climate warming can also cause a change in conditions that would set the grounds for a shift to a new stable state (Mann et al. 2012).

The deciduous dominated forest has a competitive advantage in warm and dry conditions with shallow soil organic layers. Again, once the required shallow layer is present, it has a strong reinforcing effect on the deciduous forest. A shallow soil organic layer maintains warm and dry soil conditions that speed up the decomposition of organic litter, maintaining the required depth of soil organic layer (Johnstone et al. 2012).

Shift from deciduous-dominated  to coniferous forest 

In order to shift from regime 2 to regime 1, a prolonged climatic cooling would be necessary. A slow change of the environmental conditions towards those that favour a coniferous boreal forest could, if the forest is left undisturbed for a sufficiently long time, lead to late successional coniferous species, like the black spruce, replacing early successional deciduous species (Rupp et al. 2002). Although the spatial scale considered is of extreme importance, sufficiently large external shocks, such as deciduous-specific insect outbreaks, could clear sufficiently large areas to advance the dominance of coniferous trees. In such hypothetical events, coniferous trees could get the opportunity to sprout and start altering the soil conditions to their own favour.

The provisioning of wild products such as berries and game is expected to shift some species and decrease others, but it is unclear what the net changes will be (Chapin et al. 2008). The boreal forest is likely to become a net source of carbon to the atmosphere (McGuire et al. 2009; Schuur et al. 2009), while the air quality is expected to decrease periodically due to the increase in fire frequency (Chapin et al. 2008).

The primary tactic used for reducing impacts of fire and enhance resilience (i.e. maintain structures, feedbacks, functions and processes of regime 1 is fire suppression to reduce the fire severity (Chapin et al. 2008). Decrease in fire severity could prevent the fire from consuming the soil organic layer, which in turn favours the conditions for coniferous tree. Fire suppression should be maintained in Alaska since it minimizes the risk fire imposes to life and property in communities. However, distant fires should be allowed to burn under controlled conditions, since it could have a cooling effect at a regional scale due to the higher albedo of post-fire deciduous species compared to the previous coniferous forest (Chapin et al. 2008).

Reducing air temperature is essential to be able to restore regime 1. Alaska itself accounts for a miniscule proportion of global greenhouse gas emissions and consequently to climate warming (Chapin et al. 2008). Therefore, reducing Alaska’s emissions will not have a major impact on global climate change. Global abatement of emissions is the long-term solution. Regardless of global greenhouse gas emission policies, it is highly likely that recent warming and wildfire frequency will continue for several decades in Alaska due to the multidecadal lag in the climate system (IPCC, 2007). This makes the reduction of emissions an inefficient short-term solution.  Secondary challenges that face efforts to achieve reduced emissions include restricting the economy and changing the focus of developed nations on a continued economic but sustainable growth. Educating global public about the social impacts of climate change in Alaska and other parts of the world (e.g. reduction in ecosystem services) may promote the willingness to reduce emissions (Chapin et al. 2008).

In summary, this means that the regime shift from coniferous to deciduous dominated forest in theory could be reversible through manipulation of e.g. soil conditions or local climate. However, due to projected global socio-economic development and climate change, it is highly unlikely that reversing the regime shift will be possible during the next 100 years.

This regime shift involves an abrupt reorganization of an aquatic food web due to a decrease in top predators or increase in environmental factors like temperature. Trophic cascades play a central role in these regime shifts. During these cascades groups in adjacent trophic levels often show inverse patterns in their abundance (Carpenter 2003). In other words, every other trophic level displays an increase in biomass, which leads to increased predation pressure and biomass decrease on every other trophic level. Most trophic cascades are described in ecosystems with low species diversity and/or simple food webs and/or small size (Frank et al. 2005). The magnitude of a trophic cascade varies depending on species diversity, regional oceanography, local physical disturbance, habitat complexity and fishery practices (Salomon et al. 2010). Food webs with more trophic levels and biodiversity show less trophic cascades than simpler systems (Salomon et al. 2010); due in part to omnivory and the variation on trophic interaction strength (Bascompte et al. 2005). However some documented examples includes subtidal reefs in New Zealand (Shears and Babcock 2002), Caribbean coral reefs (Bellwood et al. 2004), the Gulf of Maine (Steneck et al. 2004), and the North Atlantic (Kirby et al. 2009).

Predator-dominated food web.

Ecosystems in this regime are characterized by high predatory fish, low planktivorous fish, high zooplankton and low phytoplankton abundance. Such ecosystems can be subject to human influence, such as intensive fishery, but exhibit so called “natural compensation” e.g. via species richness and omnivory (Pace et al. 1999) that maintain the regime. Despite high productivity, such ecosystems are less likely to suffer from eutrophication due to high grazing pressure on phytoplankton. Ecosystem services associated with predator-dominated food webs includes food provision, higher biodiversity, better disease and pest control, as well as cultural services like recreation and aesthetic values (e.g. diving, sport fishing).

Planktivore-dominated food web.

Ecosystems in this regime are characterized by low predatory fish, high planktivorous fish, low zooplankton and high phytoplankton abundance. In extreme cases the actual number of trophic levels (TLs) can decrease. Ecosystems with lower trophic level dominance are more vulnerable to eutrophication due to lowered grazing pressure on phytoplankton, as well as invasion of planktivore organisms like jellyfish. A planktivore-dominated food web can affect the fluxes of carbon, transforming aquatic systems -coastal and lakes- from sinks to sources of green house gases (Bakun et al. 2010; Estes et al. 2011). Other ecosystem services like fisheries and recreation are expected to be significantly reduced.

Trophic cascades are usually initiated by fishing pressure that have direct or indirect impacts on the abundance of top predators. The response time to these drivers varies across ecosystems. In the case of coastal areas it has been found that the ecosystem response time to overfishing can vary from decades to centuries where multiple predators exist (Jackson et al. 2001).

Other factors that affect the regime shift are global warming and demand of food and fiber. Global warming have effects on upwelling, water circulatory systems that bring nutrients from the bottom of the sea towards the surface. By altering upwellings, food webs receive either too much or too little nutrients. In both cases the energy transferred towards higher trophic levels diminish as nutrients input variability increase. Demand of food strongly drives fishing, being the main cause of lost of top predators over the world.

Predator dominated food webs occur in environments with low disturbance both from fishing and nutrients input. They usually have four or more trophic levels. The regime is maintained by biotic mechanism that comprise predation and competition both intra and inter species. Abiotic factors also play a role maintaining the regime, by allowing an intermediate level of nutrient inputs from the deep ocean necessary for sustaining the food web.

In general, a combination of low top predator productivity due to poor recruitment or somatic growth combined with high exploitation pressure are conditions under which regime shifts in aquatic food webs are likely to occur (Daskalov et al. 2007). Overfishing can unbalance trophic relations, changing the configuration of competition and predation. Weak competitors may become stronger given less fishing pressure on such stocks, or by favoring the reproduction of it’s preys by fishing its competitor species. In addition, changes in nutrients inputs can represent an advantage for lower trophic levels, allowing them to escape the control of its predators.

As result, the ecological function of certain species groups is changed, allowing over abundance of immediate lower level. For example, the decrease of top-predators allows over abundance of meso predators that in turn reduces planktivorous fish and increase zooplankton. Further increase in nutrients from upwellings can lock the food web in a state dominated by planktivore fish as jellyfish. Low nutrients input, on the other hand, would reduce the amount of energy available for the food web and as a result an abrupt reduction of higher trophic levels is expected.

Shift from predators to planktivore dominated food webs

Commercial fish stocks that are locked into a regime where they are maintained at low levels of abundance has obvious economic consequences on fisheries industry. However, these fisheries related shifts are not always negative, as for example a decrease in top predators can increase the catches of mid-level prey (e.g. shrimps and clupeids) (Frank et al. 2005). When an ecosystem-wide regime shift occurs it affects not only targeted groups, such as commercial fish (Daskalov et al. 2007) but also impacts other ecosystem functions. Biodiversity as well as ecosystem resilience may be reduced by a shift from an ecosystem with high top predator diversity and dominance to a system with lower trophic level dominance. This means that ecosystem may become more vulnerable to both climatic and anthropogenic change. Such ecosystems may more easily suffer from eutrophication, hypoxia and invasion by non-native species. Eutrophication and hypoxia in turn lead to increased algal blooms and decrease the recreational values of water bodies.

Predators and commercial fish stocks have remained under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the reduction of food webs, which in turn increases the risk of fisheries collapse. Depletion of fish stocks has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Hassan et al. 2005). Fish catches are projected to further decrease in the 21st century affecting protein sources for people, especially in poor regions (Hassan et al. 2005). The estimated current contribution of fisheries to human protein consumption is 29 million tons produced industrially and 24 million tons in small-scale fisheries (Hassan et al. 2005).

Options for enhancing resilience

In food webs that exhibit strong top-down control, decreasing the fishing pressure on top predators is an obvious strategy (Scheffer et al. 2001; Frank et al. 2005). However, due to the hysteresis effect, return to a predator-dominated food web may often require substantially lower fishing levels than that which induced the regime shift, together with additional measures. Hence, restoration is usually costly, especially to fisherman (Moellmann et al. 2009). As mentioned above, environmental factors can play an important role in maintaining alternative food web regimes. Nutrient loading is one of the key parameters controlling processes such as eutrophication and hypoxia. However, decreasing nutrient loads in marine environments requires large scale actions and often collaboration at international level (e.g. Helsinki Commission Baltic Sea Action Plan in the Baltic Sea). Hence, the time-frame of these actions in addition to time-lag in ecosystem response is often long. Climatic variation cannot be directly controlled by human actions. However, climate impacts can be mediated by employing other management measures that contribute to maintaining healthy marine ecosystems with high biodiversity and resilience, which then can compensate for changes in climate.

Options for reducing resilience to encourage restoration or transformation

In addition, in small closed food webs systems (3 to 4 trophic levels), biomanipulation has been suggested to manage one of the symptoms of degraded marine environments, namely eutrophication. This involves increasing the population of predatory fish such as bass, pike and walleye through stocking or reduced angling quotas. Increased populations of these predators leads to a decrease in the level of zooplanktivores. This in turn allows an increase in the population of planktivores that graze on the algae and zooplanckton, helping to reduce the algal density and phytoplankton respectively(Smith and Schindler 2009). This option, however, has limited reach when it comes to open marine foodwebs, where seasonal migrations and metapopulation dynamics are common.

During the 20th century global fisheries greatly expanded in terms of effort and range. This expansion resulted in the overfishing of many fish stocks, and in some cases triggered fisheries collapses, for example the collapse of the highly productive Newfoundland cod fishery (Worm et al. 2009). While overfishing can severely deplete fish stocks, it will not necessarily trigger a fisheries collapse. A collapse occurs when once fishery pressure has been removed, stocks fail to rebuild (Hutchings 2000, Kirby et al. 2009), suggesting that the system has become locked into an alternate regime. In this case, alternate regimes are usefully conceptualized as alternative food web structures:

High abundance of a commercial fish species.

In this regime a valuable fish species is common and the fishery is highly productive. The main ecosystem services associated with high abundance of commercial fish are the production of food in first place, employment for both artisanal and industrial fishermen; followed by pest and disease regulation by healthy food webs, recreational values as recreational fishing, knowledge and educational values, as well as maintenance of biodiversity.

Low abundance of the commercial fish species.

This regime is characterized by a substantially reduced abundance of the valuable fish species. In cases where the valuable fish species performs key ecological functions, it can lead to trophic cascades that lead to an overabundance of planktivorous fish or primary producers such as seagrass, macroalgae, sponge or phytoplankton (Jackson et al. 2001). With low fish abundance, food production is the most immediate ecosystem services affected. While industrial fishermen often can move and harvest new areas or new species, artisanal fishermen often can lose their livelihoods and traditions.  

There are two key direct drivers of fisheries collapse: overfishing as a top-down disturbance, and the change in nutrients input as a bottom-up disturbance.

Overfishing reduces the population size of commercial fish species, reduces their ability to perform their functional role in the ecosystem, can destabilize their population dynamics (Anderson et al. 2008), and ultimately could change the structure of the food web. Nutrient inputs, on the other hand, affect fisheries by increasing the abundance of lower trophic levels e.g. phytoplankton and zooplanktivores such as jellyfish (Daskalov et al. 2007). Such changes in food web configuration can reduce the energy transfer to fish and hence reduce fishing productivity. These drivers can reduce the resilience of a regime making it more vulnerable to other types of regime shift, such as climate driven food web reorganization [see: Climate & Marine Food Webs]. The indirect drivers that explain overfishing and changes in nutrient input interact in synergistic ways, making their identification a challenging process (Ocean Studies Board 2006, Kirby et al. 2009).

Indirect drivers of overfishing include oversupply of fishing boats, inflexible fishing quotas, demand for fish, the development of new fishing technologies, the facilitation of trade (market connectivity), governance failures such as the failure of regulation to halt illegal fishing, and market failures such as subsidies or perverse incentives. Nutrient enrichment of marine ecosystems is also indirectly driven by runoff of fertilizer nutrients from agriculture and sewage from urban areas (Diaz and Rosenberg 2008).

A fishery collapses when a high-valued commercial species is substantially depleted and fails to immediately recover when fishing effort is removed (Hutchings 2000). It is usually associated with a substantial and persistent change in the structure of the marine community. Such a change involves changes in the species presence, their relative abundance, size and age.

When the stock of commercial species decreases or even disappears the populations of other similar species may increase. Many commercially fished species are predatory fish, such as tuna, that have a high trophic level. Removing such top predators can cause a trophic cascade to reshuffle a food web (Daskalov et al. 2007, Estes et al. 2011), especially if there are not ecologically similar species present. Such trophic cascades, can increase fish stocks of prey species (Walters and Kitchell 2001), which can in turn lead to either sequential depletion of fish stocks or the development of new fisheries (Essington et al. 2006).

The low abundance of commercial fish species can be maintained by a number of mechanisms. First, when the population of the commercial fish species falls under certain levels their population growth rate can decline from a variety of causes. This is known as depensation in fisheries, or the Allee effect more generally (Liermann and Hilborn 1997). Such an effect can occur if small schools of fish offer poorer protection from predators, smaller populations are less able to hunt or detect food, or mating success rates can decline.

Second, nutrient enrichment can favor the food web shifts towards zooplanktivores, such as jellyfish. By reducing resources for other fish, the rise of such species can suppress fisheries recovery. Fishery pressure on filter-feeding fishes like sardine exacerbates the problem by favoring jellyfish outbreaks (Jackson et al. 2001, Brotz et al. 2012, Condon et al. 2013).

Despite evidence from some well-documented cases, the degree to which fisheries declines are regime shifts remains a controversial issue for a number of reasons (Hilborn 2007, Litzow and Urban 2009). First, time series for most fisheries are only available from 1950, a time when marine ecosystems were already affected by flourishing industrial fishing, and some stocks had already been seriously depleted, especially those of large marine vertebrates like whales, manatees, crocodiles, seals, swordfish, sharks and rays (Kirby et al. 2009). This lack of long term data makes it difficult to identify causation in fisheries time series. Furthermore, marine ecological diversity and food web complexity complicates the detection of changes in fish populations due to fishing, because the signals of regime shift dynamics are small relative to ecological variation (Liermann and Hilborn 1997).

Shift from high to low abundance of commercial fish species

Since 1950 the world marine fisheries catch has expanded almost fivefold, and is currently about 80 million tonnes of fish. According to FAO, about 30% of fish stocks are overexploited, 60% are fully exploited and about 10 % are not fully exploited (FAO 2012). Global marine fish catches have plateaued and are not expected to increase. Local fisheries decline has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Pauly et al. 2006). There is not clear assessment of the number of stocks that have collapsed but FAO suggests about 5% of fish stocks are depleted (FAO 2012). Some collapsed stocks have remained at under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the collapse of fisheries.  

Shift from low to high abundance of commercial fish species

An abundant fish population regime will likely better maintain marine biodiversity and will be more resilience to climate variation and climate change. Society will benefit of high abundance regime in terms of high primary productivity that translate in more food, jobs and livelihoods in small coastal communities. It will also better regulate marine hypoxia.

A review of marine fisheries in the Millennium Ecosystem Assessment offer a synthetic set of managerial options for addressing fisheries collapse (Pauly et al. 2006).

Options for enhancing resilience

There are two types of strategies for enhancing resilience. The first is to stop doing things that are eroding resilience, while the second is to being activities that build resilience. There are a number of fishing practices that could be eliminated because they damage the resilience of fisheries. For example, fishing practices that destroy habitat such as bottom trawling should be avoided. Developing alternative technologies and reduction of fishing effort are suggested. Bycatch, the killing but not consuming of unwanted fish, can also have substantial impacts and can be reduced. Indirectly, fisheries subsidies and inflexible quotas can drive fishing effort to high levels that erode the resilience of a fishery.

Options for reducing resilience of unwanted regime to encourage restoration or transformation

Direct resilience building strategies include the establishment of Marine protected areas (MPAs) (Takashina and Mougi 2014). These areas can play an important role by providing refugia for fish and increasing ecological diversity. Research suggests that they need to cover ecologically significant areas, including different habitats, ensuring connectivity among them. However there remains controversy over the extent to which MPAs shift fishing pressure to less protected sites.

Indirect resilience building efforts include efficient regulation and policing of fishing can reduce illegal fishing, while comprehensive fisheries policies at the international and national level can also build resilience if they result in fisheries policies that reinforce one another. Indirect resilience building strategies include improving the ability of fishers to exit fisheries.

Good management of fishing pressure has been able to restore fisheries stocks. For example, in the US fishing closures and implementation of management measures for bottom fishing have enabled the restoration of George Bank haddock, Atlantic scallops, George Bank yellowtail flounder, Atlantic striped bass, Atlantic Arcadian redfish, Pacific chub mackerel, and Pacific sardine. However, there is no clear strategy that works across many ecosystems or nations (Beddington et al. 2007).

Mangroves are ecosystems adapted to the mixture of fresh and salt water common in intertidal zones of marine coastal environments. Mangroves cover between 1.7-1.8 x 105 km² globally. They develop within 30ºN and 38ºS and grow optimally between 15ºC and 25ºC (Lovelock 2008). Outside this geographical and thermal range mangroves show reduced leaf formation and above 38ºC or below freezing temperatures they do not photosynthesize (Mcleod and Salm 2006; Cavanaugh et al. 2014). Mangrove ecosystems are often used by local communities who depend on fishing and timber harvesting.

Mangrove forest

Mangrove forest is characterized by an ensemble of species with aerial roots adapted to survive in environments where water salinity and tides are highly variable. Mangroves grow in tropical and subtropical waters, where the optimal temperature range is between 15-24ºC. Different species occur under different salinity gradients and some trees reach 40m high in areas with high precipitation. Mangroves usually establish in soil enriched by the sediment discharge of rivers (Mcleod and Salm 2006). Mangroves are ecologically important because they provide habitat for many marine species at some stages of their lifespan, increasing both marine and terrestrial biodiversity.

Salt marshes, rocky tidal, shrimp farms.

Mangroves can become dominated by salt marshes or other configuration of coastal wetlands depending on the substrate, sediment deposition and the topography (Mcleod and Salm 2006). The alternative regime once a mangrove has collapsed is not straightforward. It strongly depends on the land use, topological characteristics, latitude and the history of the system disturbance (Cavanaugh et al. 2014). For example, many mangroves around the world have been converted into ponds for shrimp farming, salt extraction, or simply dry out to establish agriculture, cattle ranching, infrastructure development or human settlements.

The main drivers of mangrove collapse have typically been land use/cover change during the last 50 years. One third of world’s mangroves have been lost due to overexploitation of forest resources (deforestation) or conversion into agriculture, salt extraction, or ponds for shrimp farming (Cavanaugh et al. 2014). Other threats to mangroves include the development of infrastructure (roads, dams), the diversion of fresh water for irrigation and the development of urban areas. These impacts are typically stronger in developing countries where it is expected that mangroves will decline 25% by 2025 (Mcleod and Salm 2006), while recent studies show that mangrove areas are declining 1-2% yearly (Duke et al. 2007; Alongi 200; Cavanaugh et al. 2014).

Current climate change is also expected to affect mangroves via the increase in frequency and severity of storms, floods and droughts (climate extreme events), the increase of temperature, sea level rise, and ocean acidification. Frequent extreme events that potentially decrease the flow of fresh water will increase salinity, putting stress on some species and reducing their growing rates and seedling survival (Mcleod and Salm 2006). Precipitation increase, on the other hand, could favor mangrove over salt marshes by decreasing salinity and giving a competitive advantage to mangroves (Cavanaugh et al. 2014). However, strong storms and flooding events could prevent mangroves from respiring when their aerial roots are underwater. Although temperature is expected to increase 2-6ºC before 2100, its effect on mangroves is contested. On the one hand, a 6ºC increase would heterogeneously affect mangroves around the world, although it is likely to induce thermal stress, it won’t be strong enough to surpass the temperature tolerance of these ecosystems. An increase in temperature will increase the melting rate of ice and expand the volume of world’s oceans increasing the level of the sea. On the other hand, it has also been observed that warming on the coldest time of the year can favor mangroves over taking salt marshes in temperate areas of the planet, with a suggested threshold of -4ºC for Florida (Cavanaugh et al. 2014).

Sea level rise is perhaps the most important threat to mangroves. While the expected sea level rise projection range from 0.09 to 0.88 m (or 0.9 to 8.8 mm per year) several studies report that mangrove ability to migrate upwards range from 1.2 to 4.5mm per year (Mcleod and Salm 2006). Although faster migration has been observed in Holocene stratigraphic records for Florida, migration also depends of local topological conditions. Increase in atmospheric carbon dioxide could increase mangroves growth but also ocean acidification. While the increase in growth is not expected to be strong enough to overcome the pressure of sea level rise, acidification is expected to have an indirect effect on mangroves by reducing coral reefs accretion, increasing in turn the erosive effects of waves on mangrove soils (Mcleod and Salm 2006).

Shift from mangroves to salt marshes, rocky tidal or shrimp ponds.

By having aerial roots mangroves adapt to very specific conditions where salt and fresh water meets in coastal zones. Mangroves build up carbon rich soils also known as peat, which favors further mangrove growth.

Loss of mangrove area decreases peat production. Peat is lost by wave erosion and reduces the likelihood of recovery of mangrove forest. Mangrove area loss is mainly driven by conversion to agriculture, urban settlements, overexploitation of wood, shrimp farming, salt extraction, or infrastructure development (e.g. roads, dams, water divergence to irrigation) (Cavanaugh et al. 2014). Human perturbations are maintained by social feedbacks such as demand for food and market access, population growth and the cumulative benefit of human settlements (higher provision of services). In contrast, communities with poor services access such as electric energy or gas for cooking strongly rely on mangrove wood as source of energy and construction material (Mcleod and Salm 2006).

If sea level rise occurs faster than the ability of mangroves to build up peat or migrate to higher areas, mangrove forests will disappear in major areas of the world (Alongi 2008). Both the impact of sea level rise and mangrove’s ability to adapt depend on local conditions and are highly uncertain. For example, mangroves in carbonate substrates (such as coralline islands) and microtidal systems (short tidal change from high to low) are less likely to adapt and migrate as climate change continues. Mangroves in rich soil substrates and macrotidal systems are more likely to migrate if salinity balance and sedimentation do not overcome their tolerance range (Mcleod and Salm 2006).  

Shift from salt marshes to mangroves  

Mangroves have also been observed to migrate pole wards colonizing habitats previously dominated by salt marshes (Cavanaugh et al. 2014). The shift occurs when the minimum temperature in winter increase for a longer period of time, giving mangroves an advantage to over compete salt marshes. It has been observed in areas such as Florida, Louisiana or Australia. Although there is uncertainty regarding the mechanism, the threshold is thought to be related with the number of days that minimum temperatures are colder than -4ºC, at least for Florida case (Cavanaugh et al. 2014). The authors of the Florida study also report that it’s unlikely that local scale drivers such as nutrients inputs, sedimentation or hydrology have had a determinant effect on the shift. However, sea level rise has contributed to increasing mangrove area (since it has happen slowly enough) and other common driver of the shift can reduce mangrove accretion such as coastal development, aquaculture and timber production (Cavanaugh et al. 2014).

Mangroves provide a variety of ecosystem services. Mangrove ecosystems provide habitat for economically important marine species for fishing (e.g. shrimp or lobster among other crustaceans and mollusc), and they provide habitats for important groups of terrestrial species such as reptiles, monkeys, and birds. They are therefore important for maintaining both marine and terrestrial biodiversity. Mangroves also provide fuelwood, charcoal, and wood for construction. As a forest it provides protection against storms, floods, river born siltation and also traps water pollutants and sediments, reducing the erosive action of waves, particularly protecting adjacent ecosystems such as coral reefs, and sea grass beds (Duke et al. 2007). Mangroves have also been shown to play a key role in the global carbon budget, capturing up to 15% of global carbon and exporting up to 11% of particular terrestrial carbon to the oceans (Alongi 2014). In fact, recent studies support the hypothesis that mangroves are able to capture and store carbon in a greater extend that terrestrial ecosystems (Lovelock 2008), nearly identical to those of tropical forest (Alongi 2014). Therefore, losing mangrove area implies the loss of carbon storage.

While in the short term local communities could benefit by converting mangroves into shrimp farms, agriculture fields, or using its wood; in the long term they mean the loss of important services such as coastline protection, biodiversity important for tourism and fishing, water cleansing and carbon storage (Ewel et al. 1998; Cavanaugh et al. 2014). In fact, their services has been valued over $1.6 trillion per year (Costanza et al. 1997).

McLeod et al. (2006) propose a series of tools to monitor and plan for mangroves adaptation to climate change. First they propose to assess mangrove vulnerability based on the local conditions. Management options for mangrove forests are highly context dependent. As the main threat to mangroves is climate change through sea level rise, mangrove ecosystems will likely migrate upwards and northwards (Cavanaugh et al. 2014). However, such migration is constrained by local conditions such as substrate types, infrastructure development, sediment input and changes in salinity.

Second, they propose to apply risk-spreading principles (Duke et al. 2007) and identify areas that are likely to be sources and sinks of the migratory process (Mcleod and Salm 2006).  By identifying the role of each mangrove patch, it is easier to prioritize where green belts are needed to increase connectivity, which patches should be protected and which patches are more likely to respond to restoration efforts. They emphasize the importance of establishing a baseline and monitoring program with local communities in order to assess the main drivers locally. Developing partnerships with local stakeholders and developing alternative livelihoods for mangrove dependent communities are key managerial efforts (Mcleod and Salm 2006). The resilience of mangroves to future climate change scenarios also depend on how successful management efforts are at reducing the influence of other drivers: deforestation, overexploitation, infrastructure development, divergence of water for irrigation, and urbanization. Management of such local forcing will increase resilience to regional forcing such as increasing storm events or ocean acidification.

Third, it has been suggested that policy and market-based are critical to foster conservation efforts, reduce mangrove encroachment and maintain carbon storage through funding mechanisms such as REDD or other payment for ecosystem services schemes (Hutchison et al. 2013). Effective governance structures and education has also been suggested as managerial strategies (Duke et al. 2007).

Co-operation and sustainable resource levels

Harvesters extract from the resource at rates that are socially optimal, ensuring that resource levels stay at their most productive level and the community-average payoff is high. Social capital exists within the community, ensuring any defectors are ostracised.

Over-harvesting

Harvesters exert high efforts in extracting the resource, leading to depletion of the resource and low payoffs for the community. Social capital is absent and defectors are not ostracised.

Shift from ‘Co-operation and sustainable resource levels’ to ‘Over-harvesting’

The regime shift occurs when ostracism ceases to be an effective mechanism for encouraging defectors to co-operate, because the benefits of overharvesting begin to outweigh the disadvantages of ostracism. A number of factors could drive this shift. Increasing resource level, for example due to increased inflow, can lead to ineffective ostracism because at higher resource levels defection becomes more attractive by providing higher gains from resource over-extraction. Increasing defector payoff compared to ostracism, for example due to decreased costs or decreased ostracism strength, could also lead to defection becoming increasingly attractive.

Shift from ‘Over-harvesting’ to ‘Co-operation and sustainable resource levels’

Reversal of any of the above trends can cause a shift from over-harvesting to co-operation: decrease in resource levels, for example due to decreased inflow; increased costs; or increased ostracism strength.

Co-operation and sustainable resource levels occur when there is sufficient social capital to ostracise defectors and thereby make defection less attractive than co-operation. The key negative feedback that keeps the regime stable is the following: if the fraction of cooperators increases the social capital and hence strength of ostracism increases which reduces the utility of defectors, making defection less attractive. At the same time, however, resource productivity increases which increases the utility of defectors, so stability ultimately depends on the respective strength of the two feedback loops.

The over-harvesting regime exists when there are only few cooperators and hence there is insufficient social capital to ostracise defectors. The feedback that maintains full defection is as follows: any agent that 'co-operates' (i.e. harvests at a lower level) will have a payoff substantially less than the defectors. In the absence of social capital to encourage co-operation through ostracism, the co-operator will switch back to defection. The key threshold for a collapse in co-operation is when the costs of ostracism are lower than the benefits of overharvesting, i.e. ostracism ceases to be an effective mechanism to discourage defection. A variety of drivers can cause this shift, as discussed below.

To trigger a transition from full defection back to co-operation, co-operation must become more attractive than defection. This could be achieved by increasing the costs of harvesting, or by decreasing the level of the resource to low levels so that the benefits of overharvesting are minimal. However, once the co-operation strategy is lost in the community, it may be very difficult for it to re-emerge.

The ability of the natural system to provide resources for harvesting is lost with the regime shift from co-operation to over-harvesting (gained for over-harvesting to co-operation). Depending on context, loss of other ecosystem services many accompany the decline in resources. The income that members of the harvesting community obtain by harvesting is severely decreased by this regime shift. Income may even no longer exceed the costs of harvesting and the community may need to find other means of survival.

Options for preventing regime shift to over-harvesting

Management actions that stop the drivers discussed above reaching their thresholds may help to prevent regime shifts. Activities that strengthen social norms and trust in the community and thus enhance cooperation and decrease the incentive to defect and overharvest for the individual benefit (hence increasing the strength of the ostracism).

Options for restoration of co-operation:

Introducing measures that build trust in the community and disincentivise free riding. 

Oligotrophic regime

Coastal regions serve as an attraction to people, and contribute to marine resources, recreation and transport, which on the other hand cause high human impact on these ecosystems. An oligotrophic coastal system has clear water and often submerged vegetation. Commercially preferred fish species may be abundant, because coastal ecosystems support many fisheries and are highly productive due to the nutrients from the land via runoff, or deep ocean via upwelling (Boesch 2002). Many seas experience some natural hypoxia/anoxia and eutrophication, but in smaller extent and shorter time period than in the eutrophic regime. The limited amount of nutrients keeps the growth of primary production and algal blooms restricted.

Eutrophic regime

Estuaries and coastal marine ecosystem receive anthropogenic pollutants rapidly from rivers and streams from drainage basins. In eutrophic marine ecosystems phytoplankton and epiphytic algae biomass, as well as nuisance blooms of gelatinous zooplankton, have increased. Changes in species composition can take place in all trophic levels. Phytoplankton species may have shifted to taxa that are toxic or inedible. Animal species composition often changes to include less recreationally and commercially desired species. Macroalgal and vascular plants can experience changes in biomass and species composition or disappear completely. (Smith et al. 1999; Smith 2003)

Eutrophic regime causes economic losses due to restoration and damaged goods and health threats to humans exposed to algal toxins. Water clarity is reduced and humans perceive the water overall as less aesthetic.

Eutrophication commonly takes place in marine coastal waters (Smith et al. 1999; Cloern 2001). The primary cause of marine eutrophication is excessive increase in nutrient concentration (Nixon 1995) from riverine loads, originating from fertilized agricultural areas, urban sewage and industrial wastewaters (e.g. Bonsdorff et al. 1997). The nutrient loads from land to sea have been successively increased as a result of land clearing, population growth, industrial development, increased use of fossil fuels and increased use of fertilizers in agriculture (Boesch 2002). In addition, oceanic upwelling transports nutrient-rich waters to water surface and atmospheric nitrogen can enter the ocean (Boesch 2002; Paerl 1997). Estuaries may be naturally eutrophic as the nutrient loads from land are concentrated in confined channels, but cultural eutrophication often increases the nutrients significantly. In marine systems nitrogen is usually the key limiting nutrient (Borysova et al. 2005), but the limiting nutrient might differ between locations.

Also other human impacts contribute to the changes observed in many coastal areas (Cloern 2001), such as fishing (removal of top-predators causing food web reorganizations). For instance in the Black Sea the intense eutrophication is suggested to result from the combination of nutrient inputs, low grazing pressure on phytoplankton and favorable climatic influences (Llope et al. 2011). Climate change is projected to intensify eutrophication by potentially increasing the water temperature and thereby strengthening the vertical stratification potentially leading to larger deep anoxic water volumes or by increasing the discharge due to higher rainfalls (Justic et al. 2009). Local characteristics of the sea, such as water residence time, stratification and tides affect the intensity of eutrophication and the ecosystem vulnerability to it (Cloern 2001; Justic et al. 2009).

The biological growth and algal blooms in an oligotrophic sea are restricted by the limited availability of nutrients, in particular nitrogen and phosphorus. Diverse processes tend to regulate algal growth in coastal environments, for instance herbivores control algal biomass and elevated oxygen concentrations suppress recycling of nitrogen and phosphorus (Kemp et al. 2005).

In summary, nutrient addition increases algal biomass and after substantial nutrient loading affects bottom-water oxygen and nutrient recycling, water clarity and benthic primary production (Kemp et al. 2005). Biomass increases lead to a high remineralisation of organic matter at the sea bottom leading to a drecrease in oxygen concentration (Borysova et al. 2005). Oxygen within sulphates is used by bacteria, resulting in the release of sulfur that will capture the free oxygen still available in the upper layers. Sulfur also causes the specific smell of sediments in eutrophic water. As the deep water oxygen concentration decrease towards anoxic conditions, benthic orgranisms will decrease and eventually disappear.

Positive feedback loops are found in nutrient recycling and water clarity. In the nutrient feedback loop, increase in nitrogen and phosphorus increase the amount of algae and turbidity, which decreases oxygen concentration in deep water, leading to lower redox and denitrification processes. This causes increased recycling of nutrients (ammonium, phosphate), and increases the nutrient availability causing a higher phytoplankton biomass. In the coastal water clarity feedback the increased turbidity and abundant phytoplankton leading to less light and benthic production, which results in less nitrogen and phosphorus uptake and increased resuspension, which inhibit the growth of seagrass once again means more algae and turbidity. (Kemp et al. 2005, Nyström et al. 2012).

In eutrophic systems nutrient availability regulates primary production levels. The increased occurrence of harmful algal blooms are of great concern in marine systems (Smith et al. 1999). Provisioning ecosystem services are lost when toxins cause mortality of both wild and farmed fish as well as shellfish, and other animals. The composition and structure of food webs may be altered (Caddy 1993; Smith et al. 2006), potentially affecting the resilience of the coastal ecosystem. Damage and destruction of habitats (e.g. Kemp et al. 2005; Walker et al. 2001; Andersen & Rydberg 1988) and weakened quality of nursery and spawning grounds (Borysova et al. 2005) cause commercial fish species to migrate away or die.

Over half of the world population lives within coastal areas and most of the world's fisheries are connected to estuaries and near-shore habitats (Smith 2003 ref. Hobbie 2000), making the marine eutrophication a serious problem. Humans experience illness or even mortality due to contaminated fish and seafood and harmful algal blooms. Economic losses are caused by water quality problems, expenditures undertaken to reduce or avoid the damaging effects, damage on market goods and decreased tourism (Committee on the Causes and Management of Eutrophication, Ocean Studies Board, Water Science and Technology Board 2000; Cloern 2001).

Options for preventing the regime shift

Returning to the pristine, oligotrophic state is seldom a realistic option especially with continuation of anthropogenic nutrient inputs (Boesch 2002) and the human population growth. Eutrophication can be avoided, or restored, by defining a good ecological status for example in the European coasts through the implemented Water Framework Directive (WFD), for the system and the standards (nutrient reductions) needed to achieve it (Boesch 2002). The effect of current regulations to meet these standards is estimated and additional means to satisfy the established ecological goals identified, as was also done by the first assessment of the success of the WFD. Overall, the management strategy should be integrated to include for instance the catchment basin and atmospheric impact. The use of nutrients can be reduced already in the point of release and pollution should be removed before they end off at the sea. Ecosystem based fisheries management could improve the resilience of food-web against perturbations (Llope et al. 2011)

Options for restoration of desirable regimes

The key action to decrease eutrophication is limiting (Smith 2003; Smith et al. 2006; Conley et al. 2009) and most measures needs to be taken in the surrounding catchments. Commitments and organized efforts have been made in several regions to reverse eutrophication. Examples of institutional arrangements are regulations to reduce emissions, multinational directives for the national level such as the WFD, coastal management programs and educational efforts (Boesch 2002).

Different ecosystems need separate, specific management frameworks including several sectors. Local to regional nutrient reduction programs are formed in combination of scientific knowledge and politics. Reductions can be done through wastewater treatment, emission regulations, increased efficiency of nutrient use in agriculture, reduced fertilizer use, manure management, enhancement of nutrient sinks and reduction of urban runoff (Boesch 2002).

Estimation of progress requires observations, continuous monitoring of the nutrient sources and tracking the changes in the sea back to the particular action. A time lag of years may occur before the management effects can be seen in the sea, in particular as internal phosphorus loading from sediments continues.  Modeling enables forecasting the future consequences for nutrient reduction for, for instance, water clarity, algal blooms and oxygen levels. (Boesch 2002)

Steppe

The steppe vegetation coverage is dominated by grasses growing in dry soils (Eroglu et al. 2012). In this regime, herbivores increase nutrient availability by accelerating nutrient cycling. This increases primary production. During the Pleistocene epoch, arctic steppes sustained large grazing herbivores such as mammoths, bison and yaks (Zimov 2005). Today, grasses still sustain reindeer/caribou population. Currently, the steppe communities of Yakutia, Siberia are confined to the terraces above floodplains and south facing slopes of the river valleys (Yurtzev 1982).

Tundra

The tundra vegetation coverage is dominated by mosses and shrub such as dwarf birches and willows growing in waterlogged soils. The ecosystem has lower primary production and nutrient cycling thus the growth rate is slow. Moss growth is limited by moisture so the moist and nutrient low soil promote and sustain moss growth.

Herbivory has caused large-scale vegetation changes across the Arctic (Speed et al. 2009; Bråthen et al. 2007; van der Wal 2005; Srivastava & Jefferies 1996). Herbivores trample the vegetation and deposit faeces to nutrient limited soils which promotes grass growth sustaining herbivore population. In the absence of herbivores grass litter increases causing soil moisture to increase which reduces nutrient availability promoting the growth of unpalatable mosses.

Climate can act as a driver by changing air temperatures which is associated with changes in soil moisture. Soil moisture is key component in sustaining moss or grass coverage. Moss growth is more limited by water than by nutrients. However, nutrients are more available in drier soils which promote grass growth. Currently, climate is changing, in the long term whether steppe, tundra or another regime will dominate Arctic ecosystems in uncertain.

Shift from Steppe to Tundra
Grass-dominated steppe is maintained by trampling and nutrient inputs from herbivore faeces. The enriched soils promote the growth of grass coverage (Shaver et al. 1986). Grasses have a strong influence on soil moisture as they dry out the soil through high transpiration (Zimov 2005). Dry soil conditions are associated with increased nutrient availability intensifying grass growth (Nadelhoffer et al. 1991).

In the absence of herbivores, moss coverage increases creating water-logged soils (Zimov 1995). Water-logged soils prevent decomposition of organic matter, so nutrient availability decreases, which limits the growth of grasses and, hence, promotes moss growth instead (Zimov 1995). As moss growth increases, herbivore density is further reduced because moss have lower nutritional value.

Moss-dominated tundra is susceptible to trampling, a disturbance caused by large herbivores; therefore, moss coverage increases with low herbivore density (van der Wal 2001). Mosses causes an increase in soil moisture. The combination of high soil moisture creates the conditions for water-logged soils establishing optimal conditions for moss growth (Zimov 1995).

Shift from Tundra to Steppe

Climate is the main driver that could shift tundra back to steppe; this is because large grazers don't exist any more and arctic ecosystems are vulnerable to climate change. Rising of air temperatures would deepen the thawing layer of permafrost and increase evaporation. Therefore, frozen nutrients from soil would become available, which would promote grass growth; in addition to that, higher temperatures will enhance organic matter decomposition and hence nutrient availability. Then, grasses will increase evaporation. Nevertheless, moist soil conditions might be resilient since permafrost melting will make frozen water available. If there are no draining conditions, moisture will provide a suitable environment for mosses that will insulate soil temperatures and then buffer permafrost melting.

Rising of temperatures will create dynamics on arctic ecosystems that are difficult to predict. However, recent evidence shows that tundra could turn into different types of vegetation (e.g. shrub land, forest, lake, grassland) depending on local conditions (Karlsson et al. 2011). Steppe-grasses will prevail on drier areas, but more probably as patches within a mosaic since large herbivores that maintained steppe conditions during the Pleistocene, are not present.

Shift from Steppe to tundra

Shifting from steppe to tundra had a great impact on provision of food. The thermal insulation of the tundra prevents permafrost melting, which prevents destabilization and collapse of infrastructure (Schaefer 2012). Increasing air temperature due to climate change deepens the active layer of permafrost. However, soils covered by moss, permafrost is less susceptible to degradation (Ivanova 2003), this ecosystem service is essential since permafrost contains almost twice as much carbon as the atmosphere today and its effects would be irreversible at human time scale (Schaefer et al. 2012).

The wellbeing of nomadic indigenous peoples such as Nenets, Enets, Sami, Nganasans and Selkups, are connected to Arctic that provide opportunities for grazing and food sources, as well as ecosystems that co-produce cultural ecosystem services i.e. production and conservation of indigenous knowledge, practices and believes, as well as production of scientific knowledge (Huntington 2013). Tourism is also a cultural ecosystem service that will be affected.

Tundra to steppe

Arctic soils covered by mosses, permafrost is less susceptible to degradation (Ivanova 2003). Transition from tundra to steppe implies loss of this buffering layer. Consequence, the following ecosystem services will be impacted: Water availability. Local and regional freshwater dynamics will change leading to a decline in the number of lakes and wetlands (Wrona et al. 2016; Schaefer 2012). Increased abundance of grassland will affect livestock production, steppe might benefit both reindeer hunters and herders. Protection from infrastructure destabilization; pipelines, railways and power lines across the arctic are built on solid ground provided by the permafrost that if degraded their stability might be lost. This might cause economic costs and serious ecological damages (Schaefer, 2012). Fire regulation; as moss tundra shifts to a drier regime, possibilities of wild fires will increase, which contributes to thermal erosion of permafrost (Schaefer, 2012). Prevention of climate change; permafrost degradation turns permafrost's carbon stock into a carbon source, which would accelerate climate change (Schuur et al. 2015). Loss of this ecosystem service impacts the global level, since permafrost contains twice as much carbon as the atmosphere and effects of release is irreversible at human time scale (Schaefer, 2012).

Cultural ecosystem services will also be affected; conservation of indigenous knowledge, practices and believes related to tundra would be lost, affecting the well being of t he Nenets, Enets, Sami, Nganasans and Selkups (Huntington 2013).

To prevent the steppe from shifting into tundra, greenhouse gas emissions must be reduced in order to avoid the effects of climate change. The planetary boundary value for atmospheric CO2 concentration is 350 ppm above pre-industrial level (Rockstrom et al. 2009). This was further supported by the COP21 agreement, a treaty signed to limit temperature increase to 1.5^oC compared to pre-industrial levels. These efforts are not enough to prevent the feedback loop, eliminating emission is the only solution. In the Artic, land management strategies can address feedback loop on the local level. In a study, by Post and Pederson (2008), a 5-year experimental investigation of Arctic plant community response to warming, showed that warming under continuous grazing pressure from muskoxen did not differ from plant composition without warming suggest that management of large herbivores may be an important aspect for mitigating ecosystem response to future climate change

Attempts are being made today to restore the Pleistocene steppe ecosystem on a limited spatial extent in Siberia. Sergey Zimov's Pleistonce Park, 160 km², attempts to reestablish the steppe ecosystem by reintroducing megafauna to Northern Siberia (Republic of Yakutia). Zimov (2005) and colleagues tried to test if the 'key-herbivore´ hypothesis can be verified, as today's Holocene climate should be optimal for the Pleistocene steppe vegetation. The megafauna consisting of reindeer, moose, Yakutian horses, musk oxen and bison would influence the vegetation and soil composition, by trampling on grassland and returning nutrients to the soil through their manure (Zimov 2005). The grass root systems stabilize the soil and trampling reduces the albedo, exposing ground to colder temperatures, both would prevent permafrost from melting.

Due to thermokarst processes such as permafrost melting, numerous areas of the Arctic landscape are extremely lake-rich.  The water balance of those lakes may either be maintained due to impermeable permafrost that limits groundwater flow (Marsh 2009) or lost due to seasonal drainage, a natural characteristic of the Arctic system. However, a widespread, unnatural decline in lake abundance is being observed within the Arctic due to decades of rising atmospheric temperatures in the region that has been deteriorating the state of the permafrost (Smith et al. 2005). The lake-rich landscape of the Arctic is therefore slowly transforming into a drained, dry landscape increasingly dominated by vegetation such as shrubs and graminoids (Hinzman 2005).

Thermokarst lake ecosystem

The thermokarst lake-dominated landscape is formed as a response to some imposed disturbance, such as road construction, wildfire or climatic warming, which causes permafrost to thaw, creating irregular surface topography depressions, called thermokarsts. Those may appear as near-surface massive ice melts, allowing the surface to subside (Hinzman et al. 2005). As permafrost melts over time, those depressions become filled with melted water forming the lakes characteristic of the Arctic region.

Terrestrial ecosystem

Driven by external and internal factors, Arctic lakes are drying up at an unnatural speed within the region. This shift is accompanied by changes in vegetation, for example, increased shrub- and graminord abundance and a decrease in the cover of mosses and lichens through a broad expanse of the Arctic (Hinzman et al. 2005). The changes in vegetation may then further transform the landscape through sequential effects on foraging mammals and birds, as well as aquatic fauna and insects (Hinzman et al 2005).

Shift from thermokarst lakes to terrestrial ecosystem

The main external driver of this regime shift is climate change, which is caused by elevated greenhouse gas emissions (IPCC 2007). Increased temperatures warm the soil and warms the permafrost. As the permafrost thaws, greenhouse gasses such as CO² and CH₄ that are trapped in the ice are released into the atmosphere, further increasing the effects of climate change (Hinzman 2005; Karlsson et al. 2011). At a local scale, the effects of climate change are incremental. However, as this interaction is occurs over large areas, the effects are aggregated and self-reinforcing, which is significant since permafrost and seasonally frozen grounds store about 25% of the total global soil carbon stock (IPCC 2007).

At a local scale, initial thawing of the permafrost from increased atmospheric temperatures creates a dynamic landscape of depressions (thermokarst) in which lakes are formed (IPCC 2007). As permafrost become more degraded, drainage through outlet channels occurs. Since the temperature of melted water is higher than that of ice, it will exacerbate the melting of permafrost, causing another reinforcing feedback. Once the layer of frozen ground is completely penetrated the lakes can become permanently drained allowing for vegetation to be established (Marsh et al. 2008).

Shift from terrestrial ecosystem to thermokarst lake ecosystem

Heightened air- and soil temperatures melt the ice-rich soils and permafrost on which forests in the Arctic region develop, changing the physical conditions for the forest growing on top of it. When the roots of the trees get flooded, the trees die and ponds and lakes eventually replace the forest (Hinzman 2005). However, it is important to note that the shift from a terrestrial ecosystem to a lake dominated one is not as frequent as the shift from lakes to terrestrial ecosystems. The reason for this are the reinforcing feedbacks as mentioned above.

The warming of soil temperatures leads to the melting of continuous permafrost and the subsequent increase in abundance of discontinuous or sporadic permafrost. Melting of permafrost causes depressions in the soil called thermokarsts. Increased precipitation in the form of snowfall may have either an insulating or cooling effect on soil temperatures depending on several variables such a timing, duration, accumulation, and melting processes of seasonal snow cover, density, structure, and thickness of seasonal snow cover etc. (Zhang 2005). Depending upon the local surface energy balance, the thawed ground may refreeze or the permafrost can continue to degrade (Hinzman et al. 2005). On sites where continuous permafrost is thawing the active layer of soil in which melted water can move deepens allowing for the formation of thermokarst lakes (Hinzman et al. 2005). In freezing seasons, if the temperature is low enough and the hydraulic parameters on site suitable, the formation of these lakes can contribute to the re-formation of permafrost soils. Ice-rich permafrost can also prevent surface water from infiltrating to deeper groundwater zones, causing surface soils to be very wet. Decreased albedo due to changes in vegetation, the extension of snow-free and ice-free periods on terrestrial and lake surfaces and abundance, and reduction in the area occupied by glaciers and continental ice sheets in high latitudes may act as a positive feedback into atmospheric warming (Hinzman et al. 2005).

On sites where there is an abundance of discontinuous permafrost the active layer also deepens allowing for increased surface area for sub-ground water movement that increases the rate of drainage. Thermokarst lakes and ponds may begin to fill or drain depending upon the direction of the hydraulic gradient beneath the lake (Hinzman et al. 2005). Therefore it is not the depth of the active layer alone that determines the rate of drainage. However, increased drainage of lakes may prevent the reformation of permafrost from lack of available water for freezing in winter months and it can be inferred through advection of heat from decreased albedo. Once these drained lake beds become permanently drained they are likely to be re-vegetated which will change the albedo and may cause the extension of snow-free and ice-free periods and may act as feedback into increasing atmospheric temperature (Hinzman et al. 2005).

The affects of the shifts from lake to terrestrial ecosystems are aggregated at a local, regional and global scale due to the large scale and frequency at which those dynamics occur. The whole regime as described in the causal loop diagram below is a re-enforcing feedback that at its current state is leading to an overall decrease in lake dominated ecosystems and a increase in terrestrial ecosystems (Smith et al. 2005).

Even though this regime shift occurs at the local scale, its consequences go beyond that scale because it reshapes regional hydrology (through changes in water balance and surface water- connection and fragmentation) and alters regional supplies of the ecosystem (Karlsson et al. 2011). This regime shift therefore affects provisioning ecosystem services, as it will decrease the availability of freshwater. The above-ground stored freshwater, which is of vital importance to migratory birds, fish, and other wildlife used by indigenous people, will shift to below-ground stored freshwater (Artic Science Journeys Radio Stories 2005). Globally, however, freshwater will become a more abundant ES due to reductions permafrost and subsequently ice sheets and glaciers resulting from the regime’s feedbacks to climate change. Nevertheless, this global increase in the freshwater ES is not a positive one: “Increased freshwater delivery to the Arctic Ocean from reductions in ice sheets and glaciers result in rising sea level (...)” (Hinzman et al. 2005). At a local scale traditional hunting and fishing practices will be impaired (Vincent et al. 2013).  Arctic freshwater systems provide important migratory routes for fish stock. Due to this regime shit, those routes would be greatly altered in connectivity among lakes and river channels, as well as in terms of their physical coupling to the coastal marine ecosystem (Vincent et al. 2012). Globally, fishing practices would suffer as well. Increased discharge of water into the Arctic Ocean due to draining results in increased nutrient and external organic matter inputs to the Arctic Ocean that affect primary processes at the base of the marine food web (Vincent et al. 2012). Climate regulation is heavily affected on a global scale through the processes of this regime shift: as it sets up a positive feedback cycle in which the release of methane and carbon dioxide through the newly exposed soil and melted permafrost feeds into global warming. The shift from lakes to terrestrial ecosystems heightens gully erosion, a type of erosion that occurs when water is channeled across unprotected land and washes away the soil along the drainage lines. As permafrost degrades further, it causes erosion and slumping of lake edges and stream channels

The Arctic lakes are of vital importance to indigenous people as they are a source of fresh drinking water and also provides important winter transport routes. These lakes not only allows indigenous people access to their traditional hunting and fishing areas, but also to transport goods to remote communities and industries such as mining centers (Vincent et al. 2013). The regime shift would also mean an increase in products extracted from woods such as timber commodities and fuel: “increased areas of tree growth in the Arctic could serve to (…) take supply more wood products and related employment, providing local and global economic benefits” (Hassol et al. 2004).

There are few management options available to maintain the thermokarst lake ecosystems due to the long time scale at which permafrost melts and the natural dynamics within this system e.g. hydraulic fluctuations or site-specific soil conditions. One management option is to continue to reduce the emissions of greenhouse gases while simultaneously investigating further opportunities for carbon offsetting schemes.

For hydroelectric reservoirs, shifting ice conditions will have both positive and negative effects, and may require adaptive changes in operating procedures, with attention to minimize negative impacts associated with ice jams and ice breakup downstream of the spillway (Vincent et al. 2013). Fisheries management plans will also need to be adapted to the changes in migration and productivity of northern fish populations with ongoing climate change (Vincent et al. 2013). These essential resources require the development of integrated freshwater management plans, which include consideration of alternate water sources as traditional supplies change in quantity or quality (Vincent et al. 2011). The Arctic will require increasing vigilance and appropriate water management strategies to avoid and minimize the impacts of changing water impacts in the future (Vincent et al. 2013).

The boreal forest in northern Interior Alaska covers an approximate area of 48 million hectares (Mann et al. 2012). Bounded by the Alaska Range to the south, the Seward Peninsula to the west and Brooks Range to the north (Wolken et al. 2011). It is characterized by large areas of gently sloping uplands and flat lowlands isolated by mountain ranges and braided rivers with broad floodplains. During the past 50 years, the mean annual temperature in the interior boreal forest of Alaska has increased by 1.3 ºC (Hartmann and Wendler, 2005) and this has been strongly associated with recent increase in fire frequency and severity (Wolken et al. 2011). The annual area burned in Interior Alaska has doubled during the last decade (Kasischke et al. 2010). The forest has been dominated by coniferous trees (mainly black spruce), which is a group of species that thrive under colder, moist conditions in a deep soil organic layer. However, due to climate warming, Interior Alaska is currently shifting towards a deciduous dominated forest (e.g. aspen and birch) that prefer warmer and drier conditions with a shallow soil organic layer (Johnstone et al. 2010).  

Low fire frequency coniferous dominated forest

Under cold and wet conditions, coniferous trees, such as the black spruce (Picea mariana), dominate the boreal forest in Interior Alaska. The coniferous trees promote the accumulation of a deep soil organic layer, cold and moist soil temperatures, and slow decomposition rates. Due to the moistness of the soil, fires are not frequent in these forests. However, when they occur, they tend to be severe due to the high flammability of the black spruce. Due to the high water content of the soil organic layer, it will only be entirely consumed by the most severe fires (Johnstone et al. 2010). As long as the entire organic layer has not been consumed, the coniferous trees have a regenerative advantage due to their seed’s ability to sprout in moist organic soils. This means that these forests tend to return to a coniferous dominated state even after a fire (Hollingsworth et al. 2013).

High fire frequency deciduous dominated forest

A deciduous dominated forest, mainly consisting of aspen and birch trees in Interior Alaska, prefers and reinforces dry and warm soil conditions. Although these forests have a high organic litter production, the fast decomposition rate and the subsequent fast nutrient turnover, leaves a shallow soil organic layer and a nutrient rich mineral soil that favours deciduous tree regeneration. The warm and dry conditions in the forest lead to a comparatively high fire frequency, but due to the low flammability of the deciduous trees, the fires rarely become very severe (Johnstone et al. 2010). However, because of the shallow pre-fire soil organic layer, the fires tend to expose the mineral soil, which favours the post-fire reproduction of deciduous trees (Hollingsworth et al. 2013).

The main external direct driver of the regime shift is climate warming, which affects the important feedbacks of the boreal forest ecosystem. Warmer and drier climatic conditions have a direct effect on the most important soil characteristics (i.e. soil temperature, moisture and depth of the soil organic layer), influencing the vegetation cover. Coniferous trees require a cold moist soil with a deep soil organic layer to successfully reproduce. Deciduous trees, on the other hand, are competitively superior under warmer and drier soil conditions with a shallow soil organic layer (Johnstone et al. 2010). Extended summer seasons and the accompanying drier conditions create a suitable environment for an increase in the annual number of natural ignitions by lightning strikes (Kasischke et al. in press). Fires have an immediate effect on the environmental conditions of the forest, affecting factors such as soil moisture and the post-fire forest community composition (Kelly et al. 2012). Once established, the trees support and emphasize the conditions that favour their own reproduction and dominance. Under colder environmental conditions, coniferous trees, such as the black spruce, produce large amounts of organic litter, which, together with the presence of permafrost, maintains a moist soil and slow decomposition rates, contributing to a deep soil organic layer. Under warmer climatic conditions, deciduous trees produce ground layer fuels that decompose at fast rates, maintaining a shallow organic layer (Johnstone et al. 2010). Global socio-economic development indirectly drives climate warming, contributing to the increased atmospheric temperatures through e.g. greenhouse gas emissions (IPCC, 2011). Fire management can have a limited effect on the regime shift through fire suppression practices.

Shift from coniferous to deciduous-dominated forest

The coniferous dominated boreal forest of Interior Alaska requires a deep soil organic layer to succeed over other classes of trees. Once the required depth is present, this has a strong reinforcing effect on the coniferous forest. This is a requisite to maintain cold moist soil conditions in which coniferous trees thrive. The cold soil temperature reduces the decomposition rate of organic litter, maintaining a deep soil organic layer (Johnstone et al. 2010).

External shocks such as severe and frequent fires or insect outbreaks might cause dramatic shifts in the soil and vegetation by impacting the soil organic layer depth and tree succession (Hollingsworth et al. 2013; McCullough et al. 1998; Wolken et al. 2011). If the severity of such external shocks is sufficient, thresholds in the most important regime characteristics will be crossed (e.g. shallow post-shock soil organic layer) creating windows of opportunity for deciduous species to dominate. Atmospheric temperature influences the frequency and severity of fires, the soil characteristics and the length of the growing season; therefore, climate warming can also cause a change in conditions that would set the grounds for a shift to a new stable state (Mann et al. 2012).

The deciduous dominated forest has a competitive advantage in warm and dry conditions with shallow soil organic layers. Again, once the required shallow layer is present, it has a strong reinforcing effect on the deciduous forest. A shallow soil organic layer maintains warm and dry soil conditions that speed up the decomposition of organic litter, maintaining the required depth of soil organic layer (Johnstone et al. 2012).

Shift from deciduous-dominated  to coniferous forest 

In order to shift from regime 2 to regime 1, a prolonged climatic cooling would be necessary. A slow change of the environmental conditions towards those that favour a coniferous boreal forest could, if the forest is left undisturbed for a sufficiently long time, lead to late successional coniferous species, like the black spruce, replacing early successional deciduous species (Rupp et al. 2002). Although the spatial scale considered is of extreme importance, sufficiently large external shocks, such as deciduous-specific insect outbreaks, could clear sufficiently large areas to advance the dominance of coniferous trees. In such hypothetical events, coniferous trees could get the opportunity to sprout and start altering the soil conditions to their own favour.

The provisioning of wild products such as berries and game is expected to shift some species and decrease others, but it is unclear what the net changes will be (Chapin et al. 2008). The boreal forest is likely to become a net source of carbon to the atmosphere (McGuire et al. 2009; Schuur et al. 2009), while the air quality is expected to decrease periodically due to the increase in fire frequency (Chapin et al. 2008).

The primary tactic used for reducing impacts of fire and enhance resilience (i.e. maintain structures, feedbacks, functions and processes of regime 1 is fire suppression to reduce the fire severity (Chapin et al. 2008). Decrease in fire severity could prevent the fire from consuming the soil organic layer, which in turn favours the conditions for coniferous tree. Fire suppression should be maintained in Alaska since it minimizes the risk fire imposes to life and property in communities. However, distant fires should be allowed to burn under controlled conditions, since it could have a cooling effect at a regional scale due to the higher albedo of post-fire deciduous species compared to the previous coniferous forest (Chapin et al. 2008).

Reducing air temperature is essential to be able to restore regime 1. Alaska itself accounts for a miniscule proportion of global greenhouse gas emissions and consequently to climate warming (Chapin et al. 2008). Therefore, reducing Alaska’s emissions will not have a major impact on global climate change. Global abatement of emissions is the long-term solution. Regardless of global greenhouse gas emission policies, it is highly likely that recent warming and wildfire frequency will continue for several decades in Alaska due to the multidecadal lag in the climate system (IPCC, 2007). This makes the reduction of emissions an inefficient short-term solution.  Secondary challenges that face efforts to achieve reduced emissions include restricting the economy and changing the focus of developed nations on a continued economic but sustainable growth. Educating global public about the social impacts of climate change in Alaska and other parts of the world (e.g. reduction in ecosystem services) may promote the willingness to reduce emissions (Chapin et al. 2008).

In summary, this means that the regime shift from coniferous to deciduous dominated forest in theory could be reversible through manipulation of e.g. soil conditions or local climate. However, due to projected global socio-economic development and climate change, it is highly unlikely that reversing the regime shift will be possible during the next 100 years.

This regime shift involves an abrupt reorganization of an aquatic food web due to a decrease in top predators or increase in environmental factors like temperature. Trophic cascades play a central role in these regime shifts. During these cascades groups in adjacent trophic levels often show inverse patterns in their abundance (Carpenter 2003). In other words, every other trophic level displays an increase in biomass, which leads to increased predation pressure and biomass decrease on every other trophic level. Most trophic cascades are described in ecosystems with low species diversity and/or simple food webs and/or small size (Frank et al. 2005). The magnitude of a trophic cascade varies depending on species diversity, regional oceanography, local physical disturbance, habitat complexity and fishery practices (Salomon et al. 2010). Food webs with more trophic levels and biodiversity show less trophic cascades than simpler systems (Salomon et al. 2010); due in part to omnivory and the variation on trophic interaction strength (Bascompte et al. 2005). However some documented examples includes subtidal reefs in New Zealand (Shears and Babcock 2002), Caribbean coral reefs (Bellwood et al. 2004), the Gulf of Maine (Steneck et al. 2004), and the North Atlantic (Kirby et al. 2009).

Predator-dominated food web.

Ecosystems in this regime are characterized by high predatory fish, low planktivorous fish, high zooplankton and low phytoplankton abundance. Such ecosystems can be subject to human influence, such as intensive fishery, but exhibit so called “natural compensation” e.g. via species richness and omnivory (Pace et al. 1999) that maintain the regime. Despite high productivity, such ecosystems are less likely to suffer from eutrophication due to high grazing pressure on phytoplankton. Ecosystem services associated with predator-dominated food webs includes food provision, higher biodiversity, better disease and pest control, as well as cultural services like recreation and aesthetic values (e.g. diving, sport fishing).

Planktivore-dominated food web.

Ecosystems in this regime are characterized by low predatory fish, high planktivorous fish, low zooplankton and high phytoplankton abundance. In extreme cases the actual number of trophic levels (TLs) can decrease. Ecosystems with lower trophic level dominance are more vulnerable to eutrophication due to lowered grazing pressure on phytoplankton, as well as invasion of planktivore organisms like jellyfish. A planktivore-dominated food web can affect the fluxes of carbon, transforming aquatic systems -coastal and lakes- from sinks to sources of green house gases (Bakun et al. 2010; Estes et al. 2011). Other ecosystem services like fisheries and recreation are expected to be significantly reduced.

Trophic cascades are usually initiated by fishing pressure that have direct or indirect impacts on the abundance of top predators. The response time to these drivers varies across ecosystems. In the case of coastal areas it has been found that the ecosystem response time to overfishing can vary from decades to centuries where multiple predators exist (Jackson et al. 2001).

Other factors that affect the regime shift are global warming and demand of food and fiber. Global warming have effects on upwelling, water circulatory systems that bring nutrients from the bottom of the sea towards the surface. By altering upwellings, food webs receive either too much or too little nutrients. In both cases the energy transferred towards higher trophic levels diminish as nutrients input variability increase. Demand of food strongly drives fishing, being the main cause of lost of top predators over the world.

Predator dominated food webs occur in environments with low disturbance both from fishing and nutrients input. They usually have four or more trophic levels. The regime is maintained by biotic mechanism that comprise predation and competition both intra and inter species. Abiotic factors also play a role maintaining the regime, by allowing an intermediate level of nutrient inputs from the deep ocean necessary for sustaining the food web.

In general, a combination of low top predator productivity due to poor recruitment or somatic growth combined with high exploitation pressure are conditions under which regime shifts in aquatic food webs are likely to occur (Daskalov et al. 2007). Overfishing can unbalance trophic relations, changing the configuration of competition and predation. Weak competitors may become stronger given less fishing pressure on such stocks, or by favoring the reproduction of it’s preys by fishing its competitor species. In addition, changes in nutrients inputs can represent an advantage for lower trophic levels, allowing them to escape the control of its predators.

As result, the ecological function of certain species groups is changed, allowing over abundance of immediate lower level. For example, the decrease of top-predators allows over abundance of meso predators that in turn reduces planktivorous fish and increase zooplankton. Further increase in nutrients from upwellings can lock the food web in a state dominated by planktivore fish as jellyfish. Low nutrients input, on the other hand, would reduce the amount of energy available for the food web and as a result an abrupt reduction of higher trophic levels is expected.

Shift from predators to planktivore dominated food webs

Commercial fish stocks that are locked into a regime where they are maintained at low levels of abundance has obvious economic consequences on fisheries industry. However, these fisheries related shifts are not always negative, as for example a decrease in top predators can increase the catches of mid-level prey (e.g. shrimps and clupeids) (Frank et al. 2005). When an ecosystem-wide regime shift occurs it affects not only targeted groups, such as commercial fish (Daskalov et al. 2007) but also impacts other ecosystem functions. Biodiversity as well as ecosystem resilience may be reduced by a shift from an ecosystem with high top predator diversity and dominance to a system with lower trophic level dominance. This means that ecosystem may become more vulnerable to both climatic and anthropogenic change. Such ecosystems may more easily suffer from eutrophication, hypoxia and invasion by non-native species. Eutrophication and hypoxia in turn lead to increased algal blooms and decrease the recreational values of water bodies.

Predators and commercial fish stocks have remained under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the reduction of food webs, which in turn increases the risk of fisheries collapse. Depletion of fish stocks has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Hassan et al. 2005). Fish catches are projected to further decrease in the 21st century affecting protein sources for people, especially in poor regions (Hassan et al. 2005). The estimated current contribution of fisheries to human protein consumption is 29 million tons produced industrially and 24 million tons in small-scale fisheries (Hassan et al. 2005).

Options for enhancing resilience

In food webs that exhibit strong top-down control, decreasing the fishing pressure on top predators is an obvious strategy (Scheffer et al. 2001; Frank et al. 2005). However, due to the hysteresis effect, return to a predator-dominated food web may often require substantially lower fishing levels than that which induced the regime shift, together with additional measures. Hence, restoration is usually costly, especially to fisherman (Moellmann et al. 2009). As mentioned above, environmental factors can play an important role in maintaining alternative food web regimes. Nutrient loading is one of the key parameters controlling processes such as eutrophication and hypoxia. However, decreasing nutrient loads in marine environments requires large scale actions and often collaboration at international level (e.g. Helsinki Commission Baltic Sea Action Plan in the Baltic Sea). Hence, the time-frame of these actions in addition to time-lag in ecosystem response is often long. Climatic variation cannot be directly controlled by human actions. However, climate impacts can be mediated by employing other management measures that contribute to maintaining healthy marine ecosystems with high biodiversity and resilience, which then can compensate for changes in climate.

Options for reducing resilience to encourage restoration or transformation

In addition, in small closed food webs systems (3 to 4 trophic levels), biomanipulation has been suggested to manage one of the symptoms of degraded marine environments, namely eutrophication. This involves increasing the population of predatory fish such as bass, pike and walleye through stocking or reduced angling quotas. Increased populations of these predators leads to a decrease in the level of zooplanktivores. This in turn allows an increase in the population of planktivores that graze on the algae and zooplanckton, helping to reduce the algal density and phytoplankton respectively(Smith and Schindler 2009). This option, however, has limited reach when it comes to open marine foodwebs, where seasonal migrations and metapopulation dynamics are common.

During the 20th century global fisheries greatly expanded in terms of effort and range. This expansion resulted in the overfishing of many fish stocks, and in some cases triggered fisheries collapses, for example the collapse of the highly productive Newfoundland cod fishery (Worm et al. 2009). While overfishing can severely deplete fish stocks, it will not necessarily trigger a fisheries collapse. A collapse occurs when once fishery pressure has been removed, stocks fail to rebuild (Hutchings 2000, Kirby et al. 2009), suggesting that the system has become locked into an alternate regime. In this case, alternate regimes are usefully conceptualized as alternative food web structures:

High abundance of a commercial fish species.

In this regime a valuable fish species is common and the fishery is highly productive. The main ecosystem services associated with high abundance of commercial fish are the production of food in first place, employment for both artisanal and industrial fishermen; followed by pest and disease regulation by healthy food webs, recreational values as recreational fishing, knowledge and educational values, as well as maintenance of biodiversity.

Low abundance of the commercial fish species.

This regime is characterized by a substantially reduced abundance of the valuable fish species. In cases where the valuable fish species performs key ecological functions, it can lead to trophic cascades that lead to an overabundance of planktivorous fish or primary producers such as seagrass, macroalgae, sponge or phytoplankton (Jackson et al. 2001). With low fish abundance, food production is the most immediate ecosystem services affected. While industrial fishermen often can move and harvest new areas or new species, artisanal fishermen often can lose their livelihoods and traditions.  

There are two key direct drivers of fisheries collapse: overfishing as a top-down disturbance, and the change in nutrients input as a bottom-up disturbance.

Overfishing reduces the population size of commercial fish species, reduces their ability to perform their functional role in the ecosystem, can destabilize their population dynamics (Anderson et al. 2008), and ultimately could change the structure of the food web. Nutrient inputs, on the other hand, affect fisheries by increasing the abundance of lower trophic levels e.g. phytoplankton and zooplanktivores such as jellyfish (Daskalov et al. 2007). Such changes in food web configuration can reduce the energy transfer to fish and hence reduce fishing productivity. These drivers can reduce the resilience of a regime making it more vulnerable to other types of regime shift, such as climate driven food web reorganization [see: Climate & Marine Food Webs]. The indirect drivers that explain overfishing and changes in nutrient input interact in synergistic ways, making their identification a challenging process (Ocean Studies Board 2006, Kirby et al. 2009).

Indirect drivers of overfishing include oversupply of fishing boats, inflexible fishing quotas, demand for fish, the development of new fishing technologies, the facilitation of trade (market connectivity), governance failures such as the failure of regulation to halt illegal fishing, and market failures such as subsidies or perverse incentives. Nutrient enrichment of marine ecosystems is also indirectly driven by runoff of fertilizer nutrients from agriculture and sewage from urban areas (Diaz and Rosenberg 2008).

A fishery collapses when a high-valued commercial species is substantially depleted and fails to immediately recover when fishing effort is removed (Hutchings 2000). It is usually associated with a substantial and persistent change in the structure of the marine community. Such a change involves changes in the species presence, their relative abundance, size and age.

When the stock of commercial species decreases or even disappears the populations of other similar species may increase. Many commercially fished species are predatory fish, such as tuna, that have a high trophic level. Removing such top predators can cause a trophic cascade to reshuffle a food web (Daskalov et al. 2007, Estes et al. 2011), especially if there are not ecologically similar species present. Such trophic cascades, can increase fish stocks of prey species (Walters and Kitchell 2001), which can in turn lead to either sequential depletion of fish stocks or the development of new fisheries (Essington et al. 2006).

The low abundance of commercial fish species can be maintained by a number of mechanisms. First, when the population of the commercial fish species falls under certain levels their population growth rate can decline from a variety of causes. This is known as depensation in fisheries, or the Allee effect more generally (Liermann and Hilborn 1997). Such an effect can occur if small schools of fish offer poorer protection from predators, smaller populations are less able to hunt or detect food, or mating success rates can decline.

Second, nutrient enrichment can favor the food web shifts towards zooplanktivores, such as jellyfish. By reducing resources for other fish, the rise of such species can suppress fisheries recovery. Fishery pressure on filter-feeding fishes like sardine exacerbates the problem by favoring jellyfish outbreaks (Jackson et al. 2001, Brotz et al. 2012, Condon et al. 2013).

Despite evidence from some well-documented cases, the degree to which fisheries declines are regime shifts remains a controversial issue for a number of reasons (Hilborn 2007, Litzow and Urban 2009). First, time series for most fisheries are only available from 1950, a time when marine ecosystems were already affected by flourishing industrial fishing, and some stocks had already been seriously depleted, especially those of large marine vertebrates like whales, manatees, crocodiles, seals, swordfish, sharks and rays (Kirby et al. 2009). This lack of long term data makes it difficult to identify causation in fisheries time series. Furthermore, marine ecological diversity and food web complexity complicates the detection of changes in fish populations due to fishing, because the signals of regime shift dynamics are small relative to ecological variation (Liermann and Hilborn 1997).

Shift from high to low abundance of commercial fish species

Since 1950 the world marine fisheries catch has expanded almost fivefold, and is currently about 80 million tonnes of fish. According to FAO, about 30% of fish stocks are overexploited, 60% are fully exploited and about 10 % are not fully exploited (FAO 2012). Global marine fish catches have plateaued and are not expected to increase. Local fisheries decline has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Pauly et al. 2006). There is not clear assessment of the number of stocks that have collapsed but FAO suggests about 5% of fish stocks are depleted (FAO 2012). Some collapsed stocks have remained at under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the collapse of fisheries.  

Shift from low to high abundance of commercial fish species

An abundant fish population regime will likely better maintain marine biodiversity and will be more resilience to climate variation and climate change. Society will benefit of high abundance regime in terms of high primary productivity that translate in more food, jobs and livelihoods in small coastal communities. It will also better regulate marine hypoxia.

A review of marine fisheries in the Millennium Ecosystem Assessment offer a synthetic set of managerial options for addressing fisheries collapse (Pauly et al. 2006).

Options for enhancing resilience

There are two types of strategies for enhancing resilience. The first is to stop doing things that are eroding resilience, while the second is to being activities that build resilience. There are a number of fishing practices that could be eliminated because they damage the resilience of fisheries. For example, fishing practices that destroy habitat such as bottom trawling should be avoided. Developing alternative technologies and reduction of fishing effort are suggested. Bycatch, the killing but not consuming of unwanted fish, can also have substantial impacts and can be reduced. Indirectly, fisheries subsidies and inflexible quotas can drive fishing effort to high levels that erode the resilience of a fishery.

Options for reducing resilience of unwanted regime to encourage restoration or transformation

Direct resilience building strategies include the establishment of Marine protected areas (MPAs) (Takashina and Mougi 2014). These areas can play an important role by providing refugia for fish and increasing ecological diversity. Research suggests that they need to cover ecologically significant areas, including different habitats, ensuring connectivity among them. However there remains controversy over the extent to which MPAs shift fishing pressure to less protected sites.

Indirect resilience building efforts include efficient regulation and policing of fishing can reduce illegal fishing, while comprehensive fisheries policies at the international and national level can also build resilience if they result in fisheries policies that reinforce one another. Indirect resilience building strategies include improving the ability of fishers to exit fisheries.

Good management of fishing pressure has been able to restore fisheries stocks. For example, in the US fishing closures and implementation of management measures for bottom fishing have enabled the restoration of George Bank haddock, Atlantic scallops, George Bank yellowtail flounder, Atlantic striped bass, Atlantic Arcadian redfish, Pacific chub mackerel, and Pacific sardine. However, there is no clear strategy that works across many ecosystems or nations (Beddington et al. 2007).

Mangroves are ecosystems adapted to the mixture of fresh and salt water common in intertidal zones of marine coastal environments. Mangroves cover between 1.7-1.8 x 105 km² globally. They develop within 30ºN and 38ºS and grow optimally between 15ºC and 25ºC (Lovelock 2008). Outside this geographical and thermal range mangroves show reduced leaf formation and above 38ºC or below freezing temperatures they do not photosynthesize (Mcleod and Salm 2006; Cavanaugh et al. 2014). Mangrove ecosystems are often used by local communities who depend on fishing and timber harvesting.

Mangrove forest

Mangrove forest is characterized by an ensemble of species with aerial roots adapted to survive in environments where water salinity and tides are highly variable. Mangroves grow in tropical and subtropical waters, where the optimal temperature range is between 15-24ºC. Different species occur under different salinity gradients and some trees reach 40m high in areas with high precipitation. Mangroves usually establish in soil enriched by the sediment discharge of rivers (Mcleod and Salm 2006). Mangroves are ecologically important because they provide habitat for many marine species at some stages of their lifespan, increasing both marine and terrestrial biodiversity.

Salt marshes, rocky tidal, shrimp farms.

Mangroves can become dominated by salt marshes or other configuration of coastal wetlands depending on the substrate, sediment deposition and the topography (Mcleod and Salm 2006). The alternative regime once a mangrove has collapsed is not straightforward. It strongly depends on the land use, topological characteristics, latitude and the history of the system disturbance (Cavanaugh et al. 2014). For example, many mangroves around the world have been converted into ponds for shrimp farming, salt extraction, or simply dry out to establish agriculture, cattle ranching, infrastructure development or human settlements.

The main drivers of mangrove collapse have typically been land use/cover change during the last 50 years. One third of world’s mangroves have been lost due to overexploitation of forest resources (deforestation) or conversion into agriculture, salt extraction, or ponds for shrimp farming (Cavanaugh et al. 2014). Other threats to mangroves include the development of infrastructure (roads, dams), the diversion of fresh water for irrigation and the development of urban areas. These impacts are typically stronger in developing countries where it is expected that mangroves will decline 25% by 2025 (Mcleod and Salm 2006), while recent studies show that mangrove areas are declining 1-2% yearly (Duke et al. 2007; Alongi 200; Cavanaugh et al. 2014).

Current climate change is also expected to affect mangroves via the increase in frequency and severity of storms, floods and droughts (climate extreme events), the increase of temperature, sea level rise, and ocean acidification. Frequent extreme events that potentially decrease the flow of fresh water will increase salinity, putting stress on some species and reducing their growing rates and seedling survival (Mcleod and Salm 2006). Precipitation increase, on the other hand, could favor mangrove over salt marshes by decreasing salinity and giving a competitive advantage to mangroves (Cavanaugh et al. 2014). However, strong storms and flooding events could prevent mangroves from respiring when their aerial roots are underwater. Although temperature is expected to increase 2-6ºC before 2100, its effect on mangroves is contested. On the one hand, a 6ºC increase would heterogeneously affect mangroves around the world, although it is likely to induce thermal stress, it won’t be strong enough to surpass the temperature tolerance of these ecosystems. An increase in temperature will increase the melting rate of ice and expand the volume of world’s oceans increasing the level of the sea. On the other hand, it has also been observed that warming on the coldest time of the year can favor mangroves over taking salt marshes in temperate areas of the planet, with a suggested threshold of -4ºC for Florida (Cavanaugh et al. 2014).

Sea level rise is perhaps the most important threat to mangroves. While the expected sea level rise projection range from 0.09 to 0.88 m (or 0.9 to 8.8 mm per year) several studies report that mangrove ability to migrate upwards range from 1.2 to 4.5mm per year (Mcleod and Salm 2006). Although faster migration has been observed in Holocene stratigraphic records for Florida, migration also depends of local topological conditions. Increase in atmospheric carbon dioxide could increase mangroves growth but also ocean acidification. While the increase in growth is not expected to be strong enough to overcome the pressure of sea level rise, acidification is expected to have an indirect effect on mangroves by reducing coral reefs accretion, increasing in turn the erosive effects of waves on mangrove soils (Mcleod and Salm 2006).

Shift from mangroves to salt marshes, rocky tidal or shrimp ponds.

By having aerial roots mangroves adapt to very specific conditions where salt and fresh water meets in coastal zones. Mangroves build up carbon rich soils also known as peat, which favors further mangrove growth.

Loss of mangrove area decreases peat production. Peat is lost by wave erosion and reduces the likelihood of recovery of mangrove forest. Mangrove area loss is mainly driven by conversion to agriculture, urban settlements, overexploitation of wood, shrimp farming, salt extraction, or infrastructure development (e.g. roads, dams, water divergence to irrigation) (Cavanaugh et al. 2014). Human perturbations are maintained by social feedbacks such as demand for food and market access, population growth and the cumulative benefit of human settlements (higher provision of services). In contrast, communities with poor services access such as electric energy or gas for cooking strongly rely on mangrove wood as source of energy and construction material (Mcleod and Salm 2006).

If sea level rise occurs faster than the ability of mangroves to build up peat or migrate to higher areas, mangrove forests will disappear in major areas of the world (Alongi 2008). Both the impact of sea level rise and mangrove’s ability to adapt depend on local conditions and are highly uncertain. For example, mangroves in carbonate substrates (such as coralline islands) and microtidal systems (short tidal change from high to low) are less likely to adapt and migrate as climate change continues. Mangroves in rich soil substrates and macrotidal systems are more likely to migrate if salinity balance and sedimentation do not overcome their tolerance range (Mcleod and Salm 2006).  

Shift from salt marshes to mangroves  

Mangroves have also been observed to migrate pole wards colonizing habitats previously dominated by salt marshes (Cavanaugh et al. 2014). The shift occurs when the minimum temperature in winter increase for a longer period of time, giving mangroves an advantage to over compete salt marshes. It has been observed in areas such as Florida, Louisiana or Australia. Although there is uncertainty regarding the mechanism, the threshold is thought to be related with the number of days that minimum temperatures are colder than -4ºC, at least for Florida case (Cavanaugh et al. 2014). The authors of the Florida study also report that it’s unlikely that local scale drivers such as nutrients inputs, sedimentation or hydrology have had a determinant effect on the shift. However, sea level rise has contributed to increasing mangrove area (since it has happen slowly enough) and other common driver of the shift can reduce mangrove accretion such as coastal development, aquaculture and timber production (Cavanaugh et al. 2014).

Mangroves provide a variety of ecosystem services. Mangrove ecosystems provide habitat for economically important marine species for fishing (e.g. shrimp or lobster among other crustaceans and mollusc), and they provide habitats for important groups of terrestrial species such as reptiles, monkeys, and birds. They are therefore important for maintaining both marine and terrestrial biodiversity. Mangroves also provide fuelwood, charcoal, and wood for construction. As a forest it provides protection against storms, floods, river born siltation and also traps water pollutants and sediments, reducing the erosive action of waves, particularly protecting adjacent ecosystems such as coral reefs, and sea grass beds (Duke et al. 2007). Mangroves have also been shown to play a key role in the global carbon budget, capturing up to 15% of global carbon and exporting up to 11% of particular terrestrial carbon to the oceans (Alongi 2014). In fact, recent studies support the hypothesis that mangroves are able to capture and store carbon in a greater extend that terrestrial ecosystems (Lovelock 2008), nearly identical to those of tropical forest (Alongi 2014). Therefore, losing mangrove area implies the loss of carbon storage.

While in the short term local communities could benefit by converting mangroves into shrimp farms, agriculture fields, or using its wood; in the long term they mean the loss of important services such as coastline protection, biodiversity important for tourism and fishing, water cleansing and carbon storage (Ewel et al. 1998; Cavanaugh et al. 2014). In fact, their services has been valued over $1.6 trillion per year (Costanza et al. 1997).

McLeod et al. (2006) propose a series of tools to monitor and plan for mangroves adaptation to climate change. First they propose to assess mangrove vulnerability based on the local conditions. Management options for mangrove forests are highly context dependent. As the main threat to mangroves is climate change through sea level rise, mangrove ecosystems will likely migrate upwards and northwards (Cavanaugh et al. 2014). However, such migration is constrained by local conditions such as substrate types, infrastructure development, sediment input and changes in salinity.

Second, they propose to apply risk-spreading principles (Duke et al. 2007) and identify areas that are likely to be sources and sinks of the migratory process (Mcleod and Salm 2006).  By identifying the role of each mangrove patch, it is easier to prioritize where green belts are needed to increase connectivity, which patches should be protected and which patches are more likely to respond to restoration efforts. They emphasize the importance of establishing a baseline and monitoring program with local communities in order to assess the main drivers locally. Developing partnerships with local stakeholders and developing alternative livelihoods for mangrove dependent communities are key managerial efforts (Mcleod and Salm 2006). The resilience of mangroves to future climate change scenarios also depend on how successful management efforts are at reducing the influence of other drivers: deforestation, overexploitation, infrastructure development, divergence of water for irrigation, and urbanization. Management of such local forcing will increase resilience to regional forcing such as increasing storm events or ocean acidification.

Third, it has been suggested that policy and market-based are critical to foster conservation efforts, reduce mangrove encroachment and maintain carbon storage through funding mechanisms such as REDD or other payment for ecosystem services schemes (Hutchison et al. 2013). Effective governance structures and education has also been suggested as managerial strategies (Duke et al. 2007).

Co-operation and sustainable resource levels

Harvesters extract from the resource at rates that are socially optimal, ensuring that resource levels stay at their most productive level and the community-average payoff is high. Social capital exists within the community, ensuring any defectors are ostracised.

Over-harvesting

Harvesters exert high efforts in extracting the resource, leading to depletion of the resource and low payoffs for the community. Social capital is absent and defectors are not ostracised.

Shift from ‘Co-operation and sustainable resource levels’ to ‘Over-harvesting’

The regime shift occurs when ostracism ceases to be an effective mechanism for encouraging defectors to co-operate, because the benefits of overharvesting begin to outweigh the disadvantages of ostracism. A number of factors could drive this shift. Increasing resource level, for example due to increased inflow, can lead to ineffective ostracism because at higher resource levels defection becomes more attractive by providing higher gains from resource over-extraction. Increasing defector payoff compared to ostracism, for example due to decreased costs or decreased ostracism strength, could also lead to defection becoming increasingly attractive.

Shift from ‘Over-harvesting’ to ‘Co-operation and sustainable resource levels’

Reversal of any of the above trends can cause a shift from over-harvesting to co-operation: decrease in resource levels, for example due to decreased inflow; increased costs; or increased ostracism strength.

Co-operation and sustainable resource levels occur when there is sufficient social capital to ostracise defectors and thereby make defection less attractive than co-operation. The key negative feedback that keeps the regime stable is the following: if the fraction of cooperators increases the social capital and hence strength of ostracism increases which reduces the utility of defectors, making defection less attractive. At the same time, however, resource productivity increases which increases the utility of defectors, so stability ultimately depends on the respective strength of the two feedback loops.

The over-harvesting regime exists when there are only few cooperators and hence there is insufficient social capital to ostracise defectors. The feedback that maintains full defection is as follows: any agent that 'co-operates' (i.e. harvests at a lower level) will have a payoff substantially less than the defectors. In the absence of social capital to encourage co-operation through ostracism, the co-operator will switch back to defection. The key threshold for a collapse in co-operation is when the costs of ostracism are lower than the benefits of overharvesting, i.e. ostracism ceases to be an effective mechanism to discourage defection. A variety of drivers can cause this shift, as discussed below.

To trigger a transition from full defection back to co-operation, co-operation must become more attractive than defection. This could be achieved by increasing the costs of harvesting, or by decreasing the level of the resource to low levels so that the benefits of overharvesting are minimal. However, once the co-operation strategy is lost in the community, it may be very difficult for it to re-emerge.

The ability of the natural system to provide resources for harvesting is lost with the regime shift from co-operation to over-harvesting (gained for over-harvesting to co-operation). Depending on context, loss of other ecosystem services many accompany the decline in resources. The income that members of the harvesting community obtain by harvesting is severely decreased by this regime shift. Income may even no longer exceed the costs of harvesting and the community may need to find other means of survival.

Options for preventing regime shift to over-harvesting

Management actions that stop the drivers discussed above reaching their thresholds may help to prevent regime shifts. Activities that strengthen social norms and trust in the community and thus enhance cooperation and decrease the incentive to defect and overharvest for the individual benefit (hence increasing the strength of the ostracism).

Options for restoration of co-operation:

Introducing measures that build trust in the community and disincentivise free riding. 

Oligotrophic regime

Coastal regions serve as an attraction to people, and contribute to marine resources, recreation and transport, which on the other hand cause high human impact on these ecosystems. An oligotrophic coastal system has clear water and often submerged vegetation. Commercially preferred fish species may be abundant, because coastal ecosystems support many fisheries and are highly productive due to the nutrients from the land via runoff, or deep ocean via upwelling (Boesch 2002). Many seas experience some natural hypoxia/anoxia and eutrophication, but in smaller extent and shorter time period than in the eutrophic regime. The limited amount of nutrients keeps the growth of primary production and algal blooms restricted.

Eutrophic regime

Estuaries and coastal marine ecosystem receive anthropogenic pollutants rapidly from rivers and streams from drainage basins. In eutrophic marine ecosystems phytoplankton and epiphytic algae biomass, as well as nuisance blooms of gelatinous zooplankton, have increased. Changes in species composition can take place in all trophic levels. Phytoplankton species may have shifted to taxa that are toxic or inedible. Animal species composition often changes to include less recreationally and commercially desired species. Macroalgal and vascular plants can experience changes in biomass and species composition or disappear completely. (Smith et al. 1999; Smith 2003)

Eutrophic regime causes economic losses due to restoration and damaged goods and health threats to humans exposed to algal toxins. Water clarity is reduced and humans perceive the water overall as less aesthetic.

Eutrophication commonly takes place in marine coastal waters (Smith et al. 1999; Cloern 2001). The primary cause of marine eutrophication is excessive increase in nutrient concentration (Nixon 1995) from riverine loads, originating from fertilized agricultural areas, urban sewage and industrial wastewaters (e.g. Bonsdorff et al. 1997). The nutrient loads from land to sea have been successively increased as a result of land clearing, population growth, industrial development, increased use of fossil fuels and increased use of fertilizers in agriculture (Boesch 2002). In addition, oceanic upwelling transports nutrient-rich waters to water surface and atmospheric nitrogen can enter the ocean (Boesch 2002; Paerl 1997). Estuaries may be naturally eutrophic as the nutrient loads from land are concentrated in confined channels, but cultural eutrophication often increases the nutrients significantly. In marine systems nitrogen is usually the key limiting nutrient (Borysova et al. 2005), but the limiting nutrient might differ between locations.

Also other human impacts contribute to the changes observed in many coastal areas (Cloern 2001), such as fishing (removal of top-predators causing food web reorganizations). For instance in the Black Sea the intense eutrophication is suggested to result from the combination of nutrient inputs, low grazing pressure on phytoplankton and favorable climatic influences (Llope et al. 2011). Climate change is projected to intensify eutrophication by potentially increasing the water temperature and thereby strengthening the vertical stratification potentially leading to larger deep anoxic water volumes or by increasing the discharge due to higher rainfalls (Justic et al. 2009). Local characteristics of the sea, such as water residence time, stratification and tides affect the intensity of eutrophication and the ecosystem vulnerability to it (Cloern 2001; Justic et al. 2009).

The biological growth and algal blooms in an oligotrophic sea are restricted by the limited availability of nutrients, in particular nitrogen and phosphorus. Diverse processes tend to regulate algal growth in coastal environments, for instance herbivores control algal biomass and elevated oxygen concentrations suppress recycling of nitrogen and phosphorus (Kemp et al. 2005).

In summary, nutrient addition increases algal biomass and after substantial nutrient loading affects bottom-water oxygen and nutrient recycling, water clarity and benthic primary production (Kemp et al. 2005). Biomass increases lead to a high remineralisation of organic matter at the sea bottom leading to a drecrease in oxygen concentration (Borysova et al. 2005). Oxygen within sulphates is used by bacteria, resulting in the release of sulfur that will capture the free oxygen still available in the upper layers. Sulfur also causes the specific smell of sediments in eutrophic water. As the deep water oxygen concentration decrease towards anoxic conditions, benthic orgranisms will decrease and eventually disappear.

Positive feedback loops are found in nutrient recycling and water clarity. In the nutrient feedback loop, increase in nitrogen and phosphorus increase the amount of algae and turbidity, which decreases oxygen concentration in deep water, leading to lower redox and denitrification processes. This causes increased recycling of nutrients (ammonium, phosphate), and increases the nutrient availability causing a higher phytoplankton biomass. In the coastal water clarity feedback the increased turbidity and abundant phytoplankton leading to less light and benthic production, which results in less nitrogen and phosphorus uptake and increased resuspension, which inhibit the growth of seagrass once again means more algae and turbidity. (Kemp et al. 2005, Nyström et al. 2012).

In eutrophic systems nutrient availability regulates primary production levels. The increased occurrence of harmful algal blooms are of great concern in marine systems (Smith et al. 1999). Provisioning ecosystem services are lost when toxins cause mortality of both wild and farmed fish as well as shellfish, and other animals. The composition and structure of food webs may be altered (Caddy 1993; Smith et al. 2006), potentially affecting the resilience of the coastal ecosystem. Damage and destruction of habitats (e.g. Kemp et al. 2005; Walker et al. 2001; Andersen & Rydberg 1988) and weakened quality of nursery and spawning grounds (Borysova et al. 2005) cause commercial fish species to migrate away or die.

Over half of the world population lives within coastal areas and most of the world's fisheries are connected to estuaries and near-shore habitats (Smith 2003 ref. Hobbie 2000), making the marine eutrophication a serious problem. Humans experience illness or even mortality due to contaminated fish and seafood and harmful algal blooms. Economic losses are caused by water quality problems, expenditures undertaken to reduce or avoid the damaging effects, damage on market goods and decreased tourism (Committee on the Causes and Management of Eutrophication, Ocean Studies Board, Water Science and Technology Board 2000; Cloern 2001).

Options for preventing the regime shift

Returning to the pristine, oligotrophic state is seldom a realistic option especially with continuation of anthropogenic nutrient inputs (Boesch 2002) and the human population growth. Eutrophication can be avoided, or restored, by defining a good ecological status for example in the European coasts through the implemented Water Framework Directive (WFD), for the system and the standards (nutrient reductions) needed to achieve it (Boesch 2002). The effect of current regulations to meet these standards is estimated and additional means to satisfy the established ecological goals identified, as was also done by the first assessment of the success of the WFD. Overall, the management strategy should be integrated to include for instance the catchment basin and atmospheric impact. The use of nutrients can be reduced already in the point of release and pollution should be removed before they end off at the sea. Ecosystem based fisheries management could improve the resilience of food-web against perturbations (Llope et al. 2011)

Options for restoration of desirable regimes

The key action to decrease eutrophication is limiting (Smith 2003; Smith et al. 2006; Conley et al. 2009) and most measures needs to be taken in the surrounding catchments. Commitments and organized efforts have been made in several regions to reverse eutrophication. Examples of institutional arrangements are regulations to reduce emissions, multinational directives for the national level such as the WFD, coastal management programs and educational efforts (Boesch 2002).

Different ecosystems need separate, specific management frameworks including several sectors. Local to regional nutrient reduction programs are formed in combination of scientific knowledge and politics. Reductions can be done through wastewater treatment, emission regulations, increased efficiency of nutrient use in agriculture, reduced fertilizer use, manure management, enhancement of nutrient sinks and reduction of urban runoff (Boesch 2002).

Estimation of progress requires observations, continuous monitoring of the nutrient sources and tracking the changes in the sea back to the particular action. A time lag of years may occur before the management effects can be seen in the sea, in particular as internal phosphorus loading from sediments continues.  Modeling enables forecasting the future consequences for nutrient reduction for, for instance, water clarity, algal blooms and oxygen levels. (Boesch 2002)

Intact WAIS (full glacial or modern interglacial)

Antarctica is divided into two large ice sheets by the Transantarctic Mountains. A large part of Antarctic ice mass consists of the WAIS, which lies on a marine sediment-draped bed many hundreds of meters below sea level (Joughin & Alley 2011; Bamber et al. 2009). Compared to the East Antarctic ice sheet, the WAIS has rapid ice flow and discharge (Anderson et al. 2002). Ice shelves, which form over the water and float permanently attached to the landmass, serve as buttresses for the inland ice streams of the ice sheets (permanent layer of ice covering land), and are vulnerable to environmental forcing. Ice volume is stable when the water entering and leaving the system are in balance; the net accumulation (mainly snow) balances losses (ice calving, basal melting at the ice sheet margins) (Ivins 2009). The stability of the marine ice sheet is created between local basal restraint of the inland ice and longitudinal stretching of the ice shelves (Joughin & Alley 2011; Ivins 2009), meaning that the marine-based bottom topography, geometry of the interior part and stability of the buttressing ice shelves (Joughin & Alley 2011) contribute to maintaining the regime of the intact WAIS. The Southern Ocean marine environment generally rich in diversity, whereas the Antarctic terrestrial environment is low in diversity and missing many taxonomic groups (Rogers et al. 2012). In this regime humans have harvested large marine mammals, fish and krill, the latter of which is in direct competition with higher trophic level species (summer breeding colonies) (Rogers et al. 2012). At the present the WAIS appears to be thinning and the extent of sea ice is increasing.

Disintegrated WAIS (extreme interglacial)

Changed internal dynamics, such as complicated ice stream flow changes or natural climatic and oceanic forcing, create instability in the WAIS. Disintegration (collapse) of the WAIS means the nonlinear process, in which the ice sheet would flow at increasing rates into the ocean as its buttressing ice shelves diminish (O'Reilly & ACOS 2013). In the absence of the WAIS, the area would be covered by a broad open sea, punctuated by islands (Scherer 1991). The changes are occurring now and the collapse of the WAIS appears to already be underway, but large uncertainty remains on the timing (Joughin et al. 2014; Joughin & Alley 2011). The absence of the WAIS would affect local ecosystems adapted to the presence of ice, and due to the sea level rise, all global coastal ecosystems and coastal human habitats. Marine fauna adapted to the sea ice dynamics would be directly impacted through habitat changes and shifts in marine isotherms, and the changes could cascade to the higher trophic levels, and alter food webs (Rogers et al. 2012; Clarke et al. 2007). The absence of WAIS would leave broad, deep seaways (Joughin & Alley 2011), increasing accessibility for fishing vessels and tourists.

Climatic and oceanic forces appear to cause the WAIS retreats. Geologic evidence shows that the Antarctic ice sheet has shrunk in the past when global temperatures were warmer, for instance during the late Quaternary period (see e.g. Barnes & Hillenbrand 2010). There have been times when parts of the WAIS have been lost to the ocean, causing raised sea levels (Ivins 2009). According to the model by Pollard and DeConto (2009), a collapse from the modern conditions could occur when sub-ice ocean melting increase from 0.1 to 2 m yr^-1 under shelf interiors, and from 5 to 10 m yr^-1 near exposed shelf edges. The mechanisms leading to the WAIS collapse are still unclear. It has been suggested that the WAIS collapse could start by the intrusion of ocean water (i.e. sea level rise) between the ice sheet and the ground. As the grounding line retreats, a strong positive feedback would be triggered when ocean water undercuts the ice sheet and causes further separation from the bedrock (Oppenheimer 1998). This feedback with sea level changes driving further retreat is contested by other studies (Alley et al. 2007; Gomez et al. 2010). At present, main cause for the thinning of the WAIS ice mass is suggested to be the intrusion of relatively warm ocean water beneath the ice. Ice shelves are very sensitive to ocean temperature changes (Turner & Overland 2009) and several studies (see e.g. Price et al. 2008; Thoma et al. 2008, Joughin et al. 2014) have shown that increased transport of warm subsurface water significantly contributes to the basal ice-shelf melt. The warm Circumpolar Deep Water (CDW) protrudes into the ice shelves through submarine troughs (Bintanja et al. 2013) and produces large melt rates at the regions where it is able to access the sub-ice-shelf cavities (Jacobs et al. 1992). Although the upper layers of Southern Ocean have cooled, the subsurface sea has in fact warmed (Robertson et al. 2002). The reasons for the warming of the Southern Ocean are not completely know. It is probably caused by multiple processes, such as increased greenhouse gases in the atmosphere, shifts in Southern Annular Mode (SAM, a high-latitude mode, also called Antarctic Oscillation) and changes in sea currents.

Other, less direct contributors to the regime shift found in literature are hydrofracturing of crevasses by surface melting water, changes in landscape (glacier behavior), wind forcing, ocean upwelling, tides and long period waves, changes in stratospheric ozone affecting atmospheric and ocean circulation, and intensification and southward migration of the Southern Ocean Westerlies due to the positive trend in the SAM (e.g. Thompson & Solomon 2002; Goosse et al. 2008; Turner et al. 2009; Joughin et al. 2014).

In the intact WAIS regime there is variability in the system depending on the natural climate variation, but the rate and scale of the changes in ice mass are not as remarkable as in the second regime, and the ice volume is maintained by equal net balance. The WAIS icescape is at the present characterized by decrease in ice mass and increased sea ice extent. The most recent research indicates that the WAIS is changing significantly and rapidly, and at an accelerating rate (O'Reilly & ACOS 2013; Abram et al. 2013; Joughin & Alley 2011; Thomas et al. 2004; Velicogna & Wahr 2006). West Antarctica is currently warming, and has been stated to be one of the most rapidly warming regions on Earth (Bromwich et al. 2012; Steig et al. 2013).

The possibility for the collapse of the WAIS has been debated in science since the 1970s (Notz 2009). It has been suggested that the WAIS retreats follow glacial and interglacial periods (Clark et al. 2002; Fairbanks 1989; Scherer 1991). The potential of the WAIS to collapse and the mechanisms leading and preventing it are still unclear, but the recent studies argue that the collapse is possible with small changes in the forces (Mercer 1978; Oppenheimer 1998; Vaughan 2008; Schoof 2007; Naish et al. 2009; Pollard & DeConto 2009; Notz 2009). A basin-scale ice-flow model by Joughin et al. (2014) provided strong evidence that the early-stage collapse has already begun. Their model strengthens the argument that the losses are melt-driven and that melt-induced ice-shelf thinning reduces buttressing, creating far greater speedup and retreat through the grounding line retreat. Ice sheet thinning appears to take place through basal melting (i.e. on the underside of the ice shelves) (Pritchard et al. 2012; Joughin & Alley 2011). The highest melting rates occur where the ice shelves interact with the warmest water. Increased atmospheric temperature and ocean temperature (possibly due to several factors such as global climate change, natural climate variation, ozone hole and changes in atmospheric and oceanic circulation) cause intrusion of warmed ocean water beneath the ice shelf of the WAIS. Relatively warm Circumpolar Deep Water (CDW) protrudes into the ice shelves through submarine troughs and produces large melt rates at the ice/ocean interface, decreasing the volume of the ice mass. The estimations of the loss rate vary both regionally and for the whole continent, see for instance King et al. (2012). Although in some regions the ice sheet is thickening, the net balance is negative (e.g. Velicogna & Wahr 2006).

The mass loss of ice is expected to increase both in volume and rate in the future when the warm circumpolar deep water is able to reach further ice shelves (Rignot et al. 2011; Hellmer et al. 2012). Unlike in the Arctic and as an unusual feature in consideration to the global warming, the extent of the Antarctic sea ice has increased in the observed period starting from 1979 (Turner & Overland 2009). Sea-ice expansion may be meltwater-induced (Bintanja et al. 2013). Meltwater from the ice shelves has lower density and thus accumulates in the top ocean layer (Price et al. 2008). This upper layer water gets fresher, and resulting cold halocline reduces the convective mixing, causing the atmosphere to cool and freeze the upper 100 m more easily (Bintanja et al. 2013). The subsurface ocean warming, mass loss of ice due to the basal melt and expanding sea ice may constitute a negative feedback loop (Bintanja et al. 2013; Zhang 2007). The fresh melt water has low density and thus accumulates in the top layer, stabilizing the ocean and resulting in less mixing between cold and warm water (Bintanja et al. 2013). The cold and fresh upper seawater layer is easier to cool by the atmosphere and freezes. This meltwater-induced sea ice expansion hinders the warm deepwater from mixing with the surface water. Another feedback, which is positive but at present contested, has been suggested to be formed by the raised sea level further undercutting the ice sheet and triggering its separation from the bedrock.

Humans mainly use the Antarctica for commercial fishing, conservation, research, and tourism. The Antarctic region also plays a remarkable role in the planetary functions. Shifting the regime to that of the fragmented WAIS would affect the regulating ecosystem services of the Antarctica, such as global climate and ocean circulation through ice and snow albedo, CO₂ uptake and icescape changes. The Southern Ocean has a rich biodiversity, and fisheries include krill and toothfish. The biogeochemical cycles of the Southern Ocean and the sea ice impact the structure and dynamics of the marine ecosystems (e.g. life cycles of organisms), especially the trophic levels that are adapted to the presence, seasonality and properties of ice (Massom & Stammerjohn 2010; Arrigo 2002; Brierley & Thomas 2002; Thomas et al. 2010; Eicken et al. 1995; Moline et al. 2008; Tynan 2010; Clarke et al. 2007; Rogers et al. 2012). Sea level rise may have severe impacts on coastal areas by submergence or increasing flooding and erosion, changing ecosystems, and increasing salinization (Nicholls et al. 2011). Fisheries management and accessibility would change due to the altered sea ice extent (Anon 2009).

The absence of the WAIS would leave broad, deep seaways (Joughin & Alley 2011) and result in forced displacement of coastal population and economy (Nicholls et al. 2011). The economic consequences depend on the time-scale: playing out over a century. The WAIS collapse would damage many coastal communities, whereas occurring over several centruries, the additional time would give possibilities to develop an appropriate risk-mitigation strategy (Dowdeswell et al. 2008; Mercer 1978; Joughin & Alley 2011). Increased vessel access to more southern locations may have implications on the safety of ocean navigation (Anon 2009). Tourism, the largest commercial activity in the Antarctic, is predicted to increase due to the extended season with reduced sea ice thickness (Anon 2009). The increased human presence might be a threat to ecosystems, for instance winter ice has usually provided a relief from the fishing pressure.

Two new studies in 2014 state that since the WAIS has reached the early stage collapse phase, the shift is unstoppable (Joughin et al. 2014, NASA in press). Joughin et al. (2014) argue that it is difficult to foresee stabilization of the system unless CDW recedes sufficienty to reduce present level of melting.

Steppe

The steppe vegetation coverage is dominated by grasses growing in dry soils (Eroglu et al. 2012). In this regime, herbivores increase nutrient availability by accelerating nutrient cycling. This increases primary production. During the Pleistocene epoch, arctic steppes sustained large grazing herbivores such as mammoths, bison and yaks (Zimov 2005). Today, grasses still sustain reindeer/caribou population. Currently, the steppe communities of Yakutia, Siberia are confined to the terraces above floodplains and south facing slopes of the river valleys (Yurtzev 1982).

Tundra

The tundra vegetation coverage is dominated by mosses and shrub such as dwarf birches and willows growing in waterlogged soils. The ecosystem has lower primary production and nutrient cycling thus the growth rate is slow. Moss growth is limited by moisture so the moist and nutrient low soil promote and sustain moss growth.

Herbivory has caused large-scale vegetation changes across the Arctic (Speed et al. 2009; Bråthen et al. 2007; van der Wal 2005; Srivastava & Jefferies 1996). Herbivores trample the vegetation and deposit faeces to nutrient limited soils which promotes grass growth sustaining herbivore population. In the absence of herbivores grass litter increases causing soil moisture to increase which reduces nutrient availability promoting the growth of unpalatable mosses.

Climate can act as a driver by changing air temperatures which is associated with changes in soil moisture. Soil moisture is key component in sustaining moss or grass coverage. Moss growth is more limited by water than by nutrients. However, nutrients are more available in drier soils which promote grass growth. Currently, climate is changing, in the long term whether steppe, tundra or another regime will dominate Arctic ecosystems in uncertain.

Shift from Steppe to Tundra
Grass-dominated steppe is maintained by trampling and nutrient inputs from herbivore faeces. The enriched soils promote the growth of grass coverage (Shaver et al. 1986). Grasses have a strong influence on soil moisture as they dry out the soil through high transpiration (Zimov 2005). Dry soil conditions are associated with increased nutrient availability intensifying grass growth (Nadelhoffer et al. 1991).

In the absence of herbivores, moss coverage increases creating water-logged soils (Zimov 1995). Water-logged soils prevent decomposition of organic matter, so nutrient availability decreases, which limits the growth of grasses and, hence, promotes moss growth instead (Zimov 1995). As moss growth increases, herbivore density is further reduced because moss have lower nutritional value.

Moss-dominated tundra is susceptible to trampling, a disturbance caused by large herbivores; therefore, moss coverage increases with low herbivore density (van der Wal 2001). Mosses causes an increase in soil moisture. The combination of high soil moisture creates the conditions for water-logged soils establishing optimal conditions for moss growth (Zimov 1995).

Shift from Tundra to Steppe

Climate is the main driver that could shift tundra back to steppe; this is because large grazers don't exist any more and arctic ecosystems are vulnerable to climate change. Rising of air temperatures would deepen the thawing layer of permafrost and increase evaporation. Therefore, frozen nutrients from soil would become available, which would promote grass growth; in addition to that, higher temperatures will enhance organic matter decomposition and hence nutrient availability. Then, grasses will increase evaporation. Nevertheless, moist soil conditions might be resilient since permafrost melting will make frozen water available. If there are no draining conditions, moisture will provide a suitable environment for mosses that will insulate soil temperatures and then buffer permafrost melting.

Rising of temperatures will create dynamics on arctic ecosystems that are difficult to predict. However, recent evidence shows that tundra could turn into different types of vegetation (e.g. shrub land, forest, lake, grassland) depending on local conditions (Karlsson et al. 2011). Steppe-grasses will prevail on drier areas, but more probably as patches within a mosaic since large herbivores that maintained steppe conditions during the Pleistocene, are not present.

Shift from Steppe to tundra

Shifting from steppe to tundra had a great impact on provision of food. The thermal insulation of the tundra prevents permafrost melting, which prevents destabilization and collapse of infrastructure (Schaefer 2012). Increasing air temperature due to climate change deepens the active layer of permafrost. However, soils covered by moss, permafrost is less susceptible to degradation (Ivanova 2003), this ecosystem service is essential since permafrost contains almost twice as much carbon as the atmosphere today and its effects would be irreversible at human time scale (Schaefer et al. 2012).

The wellbeing of nomadic indigenous peoples such as Nenets, Enets, Sami, Nganasans and Selkups, are connected to Arctic that provide opportunities for grazing and food sources, as well as ecosystems that co-produce cultural ecosystem services i.e. production and conservation of indigenous knowledge, practices and believes, as well as production of scientific knowledge (Huntington 2013). Tourism is also a cultural ecosystem service that will be affected.

Tundra to steppe

Arctic soils covered by mosses, permafrost is less susceptible to degradation (Ivanova 2003). Transition from tundra to steppe implies loss of this buffering layer. Consequence, the following ecosystem services will be impacted: Water availability. Local and regional freshwater dynamics will change leading to a decline in the number of lakes and wetlands (Wrona et al. 2016; Schaefer 2012). Increased abundance of grassland will affect livestock production, steppe might benefit both reindeer hunters and herders. Protection from infrastructure destabilization; pipelines, railways and power lines across the arctic are built on solid ground provided by the permafrost that if degraded their stability might be lost. This might cause economic costs and serious ecological damages (Schaefer, 2012). Fire regulation; as moss tundra shifts to a drier regime, possibilities of wild fires will increase, which contributes to thermal erosion of permafrost (Schaefer, 2012). Prevention of climate change; permafrost degradation turns permafrost's carbon stock into a carbon source, which would accelerate climate change (Schuur et al. 2015). Loss of this ecosystem service impacts the global level, since permafrost contains twice as much carbon as the atmosphere and effects of release is irreversible at human time scale (Schaefer, 2012).

Cultural ecosystem services will also be affected; conservation of indigenous knowledge, practices and believes related to tundra would be lost, affecting the well being of t he Nenets, Enets, Sami, Nganasans and Selkups (Huntington 2013).

To prevent the steppe from shifting into tundra, greenhouse gas emissions must be reduced in order to avoid the effects of climate change. The planetary boundary value for atmospheric CO2 concentration is 350 ppm above pre-industrial level (Rockstrom et al. 2009). This was further supported by the COP21 agreement, a treaty signed to limit temperature increase to 1.5^oC compared to pre-industrial levels. These efforts are not enough to prevent the feedback loop, eliminating emission is the only solution. In the Artic, land management strategies can address feedback loop on the local level. In a study, by Post and Pederson (2008), a 5-year experimental investigation of Arctic plant community response to warming, showed that warming under continuous grazing pressure from muskoxen did not differ from plant composition without warming suggest that management of large herbivores may be an important aspect for mitigating ecosystem response to future climate change

Attempts are being made today to restore the Pleistocene steppe ecosystem on a limited spatial extent in Siberia. Sergey Zimov's Pleistonce Park, 160 km², attempts to reestablish the steppe ecosystem by reintroducing megafauna to Northern Siberia (Republic of Yakutia). Zimov (2005) and colleagues tried to test if the 'key-herbivore´ hypothesis can be verified, as today's Holocene climate should be optimal for the Pleistocene steppe vegetation. The megafauna consisting of reindeer, moose, Yakutian horses, musk oxen and bison would influence the vegetation and soil composition, by trampling on grassland and returning nutrients to the soil through their manure (Zimov 2005). The grass root systems stabilize the soil and trampling reduces the albedo, exposing ground to colder temperatures, both would prevent permafrost from melting.

Due to thermokarst processes such as permafrost melting, numerous areas of the Arctic landscape are extremely lake-rich.  The water balance of those lakes may either be maintained due to impermeable permafrost that limits groundwater flow (Marsh 2009) or lost due to seasonal drainage, a natural characteristic of the Arctic system. However, a widespread, unnatural decline in lake abundance is being observed within the Arctic due to decades of rising atmospheric temperatures in the region that has been deteriorating the state of the permafrost (Smith et al. 2005). The lake-rich landscape of the Arctic is therefore slowly transforming into a drained, dry landscape increasingly dominated by vegetation such as shrubs and graminoids (Hinzman 2005).

Thermokarst lake ecosystem

The thermokarst lake-dominated landscape is formed as a response to some imposed disturbance, such as road construction, wildfire or climatic warming, which causes permafrost to thaw, creating irregular surface topography depressions, called thermokarsts. Those may appear as near-surface massive ice melts, allowing the surface to subside (Hinzman et al. 2005). As permafrost melts over time, those depressions become filled with melted water forming the lakes characteristic of the Arctic region.

Terrestrial ecosystem

Driven by external and internal factors, Arctic lakes are drying up at an unnatural speed within the region. This shift is accompanied by changes in vegetation, for example, increased shrub- and graminord abundance and a decrease in the cover of mosses and lichens through a broad expanse of the Arctic (Hinzman et al. 2005). The changes in vegetation may then further transform the landscape through sequential effects on foraging mammals and birds, as well as aquatic fauna and insects (Hinzman et al 2005).

Shift from thermokarst lakes to terrestrial ecosystem

The main external driver of this regime shift is climate change, which is caused by elevated greenhouse gas emissions (IPCC 2007). Increased temperatures warm the soil and warms the permafrost. As the permafrost thaws, greenhouse gasses such as CO² and CH₄ that are trapped in the ice are released into the atmosphere, further increasing the effects of climate change (Hinzman 2005; Karlsson et al. 2011). At a local scale, the effects of climate change are incremental. However, as this interaction is occurs over large areas, the effects are aggregated and self-reinforcing, which is significant since permafrost and seasonally frozen grounds store about 25% of the total global soil carbon stock (IPCC 2007).

At a local scale, initial thawing of the permafrost from increased atmospheric temperatures creates a dynamic landscape of depressions (thermokarst) in which lakes are formed (IPCC 2007). As permafrost become more degraded, drainage through outlet channels occurs. Since the temperature of melted water is higher than that of ice, it will exacerbate the melting of permafrost, causing another reinforcing feedback. Once the layer of frozen ground is completely penetrated the lakes can become permanently drained allowing for vegetation to be established (Marsh et al. 2008).

Shift from terrestrial ecosystem to thermokarst lake ecosystem

Heightened air- and soil temperatures melt the ice-rich soils and permafrost on which forests in the Arctic region develop, changing the physical conditions for the forest growing on top of it. When the roots of the trees get flooded, the trees die and ponds and lakes eventually replace the forest (Hinzman 2005). However, it is important to note that the shift from a terrestrial ecosystem to a lake dominated one is not as frequent as the shift from lakes to terrestrial ecosystems. The reason for this are the reinforcing feedbacks as mentioned above.

The warming of soil temperatures leads to the melting of continuous permafrost and the subsequent increase in abundance of discontinuous or sporadic permafrost. Melting of permafrost causes depressions in the soil called thermokarsts. Increased precipitation in the form of snowfall may have either an insulating or cooling effect on soil temperatures depending on several variables such a timing, duration, accumulation, and melting processes of seasonal snow cover, density, structure, and thickness of seasonal snow cover etc. (Zhang 2005). Depending upon the local surface energy balance, the thawed ground may refreeze or the permafrost can continue to degrade (Hinzman et al. 2005). On sites where continuous permafrost is thawing the active layer of soil in which melted water can move deepens allowing for the formation of thermokarst lakes (Hinzman et al. 2005). In freezing seasons, if the temperature is low enough and the hydraulic parameters on site suitable, the formation of these lakes can contribute to the re-formation of permafrost soils. Ice-rich permafrost can also prevent surface water from infiltrating to deeper groundwater zones, causing surface soils to be very wet. Decreased albedo due to changes in vegetation, the extension of snow-free and ice-free periods on terrestrial and lake surfaces and abundance, and reduction in the area occupied by glaciers and continental ice sheets in high latitudes may act as a positive feedback into atmospheric warming (Hinzman et al. 2005).

On sites where there is an abundance of discontinuous permafrost the active layer also deepens allowing for increased surface area for sub-ground water movement that increases the rate of drainage. Thermokarst lakes and ponds may begin to fill or drain depending upon the direction of the hydraulic gradient beneath the lake (Hinzman et al. 2005). Therefore it is not the depth of the active layer alone that determines the rate of drainage. However, increased drainage of lakes may prevent the reformation of permafrost from lack of available water for freezing in winter months and it can be inferred through advection of heat from decreased albedo. Once these drained lake beds become permanently drained they are likely to be re-vegetated which will change the albedo and may cause the extension of snow-free and ice-free periods and may act as feedback into increasing atmospheric temperature (Hinzman et al. 2005).

The affects of the shifts from lake to terrestrial ecosystems are aggregated at a local, regional and global scale due to the large scale and frequency at which those dynamics occur. The whole regime as described in the causal loop diagram below is a re-enforcing feedback that at its current state is leading to an overall decrease in lake dominated ecosystems and a increase in terrestrial ecosystems (Smith et al. 2005).

Even though this regime shift occurs at the local scale, its consequences go beyond that scale because it reshapes regional hydrology (through changes in water balance and surface water- connection and fragmentation) and alters regional supplies of the ecosystem (Karlsson et al. 2011). This regime shift therefore affects provisioning ecosystem services, as it will decrease the availability of freshwater. The above-ground stored freshwater, which is of vital importance to migratory birds, fish, and other wildlife used by indigenous people, will shift to below-ground stored freshwater (Artic Science Journeys Radio Stories 2005). Globally, however, freshwater will become a more abundant ES due to reductions permafrost and subsequently ice sheets and glaciers resulting from the regime’s feedbacks to climate change. Nevertheless, this global increase in the freshwater ES is not a positive one: “Increased freshwater delivery to the Arctic Ocean from reductions in ice sheets and glaciers result in rising sea level (...)” (Hinzman et al. 2005). At a local scale traditional hunting and fishing practices will be impaired (Vincent et al. 2013).  Arctic freshwater systems provide important migratory routes for fish stock. Due to this regime shit, those routes would be greatly altered in connectivity among lakes and river channels, as well as in terms of their physical coupling to the coastal marine ecosystem (Vincent et al. 2012). Globally, fishing practices would suffer as well. Increased discharge of water into the Arctic Ocean due to draining results in increased nutrient and external organic matter inputs to the Arctic Ocean that affect primary processes at the base of the marine food web (Vincent et al. 2012). Climate regulation is heavily affected on a global scale through the processes of this regime shift: as it sets up a positive feedback cycle in which the release of methane and carbon dioxide through the newly exposed soil and melted permafrost feeds into global warming. The shift from lakes to terrestrial ecosystems heightens gully erosion, a type of erosion that occurs when water is channeled across unprotected land and washes away the soil along the drainage lines. As permafrost degrades further, it causes erosion and slumping of lake edges and stream channels

The Arctic lakes are of vital importance to indigenous people as they are a source of fresh drinking water and also provides important winter transport routes. These lakes not only allows indigenous people access to their traditional hunting and fishing areas, but also to transport goods to remote communities and industries such as mining centers (Vincent et al. 2013). The regime shift would also mean an increase in products extracted from woods such as timber commodities and fuel: “increased areas of tree growth in the Arctic could serve to (…) take supply more wood products and related employment, providing local and global economic benefits” (Hassol et al. 2004).

There are few management options available to maintain the thermokarst lake ecosystems due to the long time scale at which permafrost melts and the natural dynamics within this system e.g. hydraulic fluctuations or site-specific soil conditions. One management option is to continue to reduce the emissions of greenhouse gases while simultaneously investigating further opportunities for carbon offsetting schemes.

For hydroelectric reservoirs, shifting ice conditions will have both positive and negative effects, and may require adaptive changes in operating procedures, with attention to minimize negative impacts associated with ice jams and ice breakup downstream of the spillway (Vincent et al. 2013). Fisheries management plans will also need to be adapted to the changes in migration and productivity of northern fish populations with ongoing climate change (Vincent et al. 2013). These essential resources require the development of integrated freshwater management plans, which include consideration of alternate water sources as traditional supplies change in quantity or quality (Vincent et al. 2011). The Arctic will require increasing vigilance and appropriate water management strategies to avoid and minimize the impacts of changing water impacts in the future (Vincent et al. 2013).

The boreal forest in northern Interior Alaska covers an approximate area of 48 million hectares (Mann et al. 2012). Bounded by the Alaska Range to the south, the Seward Peninsula to the west and Brooks Range to the north (Wolken et al. 2011). It is characterized by large areas of gently sloping uplands and flat lowlands isolated by mountain ranges and braided rivers with broad floodplains. During the past 50 years, the mean annual temperature in the interior boreal forest of Alaska has increased by 1.3 ºC (Hartmann and Wendler, 2005) and this has been strongly associated with recent increase in fire frequency and severity (Wolken et al. 2011). The annual area burned in Interior Alaska has doubled during the last decade (Kasischke et al. 2010). The forest has been dominated by coniferous trees (mainly black spruce), which is a group of species that thrive under colder, moist conditions in a deep soil organic layer. However, due to climate warming, Interior Alaska is currently shifting towards a deciduous dominated forest (e.g. aspen and birch) that prefer warmer and drier conditions with a shallow soil organic layer (Johnstone et al. 2010).  

Low fire frequency coniferous dominated forest

Under cold and wet conditions, coniferous trees, such as the black spruce (Picea mariana), dominate the boreal forest in Interior Alaska. The coniferous trees promote the accumulation of a deep soil organic layer, cold and moist soil temperatures, and slow decomposition rates. Due to the moistness of the soil, fires are not frequent in these forests. However, when they occur, they tend to be severe due to the high flammability of the black spruce. Due to the high water content of the soil organic layer, it will only be entirely consumed by the most severe fires (Johnstone et al. 2010). As long as the entire organic layer has not been consumed, the coniferous trees have a regenerative advantage due to their seed’s ability to sprout in moist organic soils. This means that these forests tend to return to a coniferous dominated state even after a fire (Hollingsworth et al. 2013).

High fire frequency deciduous dominated forest

A deciduous dominated forest, mainly consisting of aspen and birch trees in Interior Alaska, prefers and reinforces dry and warm soil conditions. Although these forests have a high organic litter production, the fast decomposition rate and the subsequent fast nutrient turnover, leaves a shallow soil organic layer and a nutrient rich mineral soil that favours deciduous tree regeneration. The warm and dry conditions in the forest lead to a comparatively high fire frequency, but due to the low flammability of the deciduous trees, the fires rarely become very severe (Johnstone et al. 2010). However, because of the shallow pre-fire soil organic layer, the fires tend to expose the mineral soil, which favours the post-fire reproduction of deciduous trees (Hollingsworth et al. 2013).

The main external direct driver of the regime shift is climate warming, which affects the important feedbacks of the boreal forest ecosystem. Warmer and drier climatic conditions have a direct effect on the most important soil characteristics (i.e. soil temperature, moisture and depth of the soil organic layer), influencing the vegetation cover. Coniferous trees require a cold moist soil with a deep soil organic layer to successfully reproduce. Deciduous trees, on the other hand, are competitively superior under warmer and drier soil conditions with a shallow soil organic layer (Johnstone et al. 2010). Extended summer seasons and the accompanying drier conditions create a suitable environment for an increase in the annual number of natural ignitions by lightning strikes (Kasischke et al. in press). Fires have an immediate effect on the environmental conditions of the forest, affecting factors such as soil moisture and the post-fire forest community composition (Kelly et al. 2012). Once established, the trees support and emphasize the conditions that favour their own reproduction and dominance. Under colder environmental conditions, coniferous trees, such as the black spruce, produce large amounts of organic litter, which, together with the presence of permafrost, maintains a moist soil and slow decomposition rates, contributing to a deep soil organic layer. Under warmer climatic conditions, deciduous trees produce ground layer fuels that decompose at fast rates, maintaining a shallow organic layer (Johnstone et al. 2010). Global socio-economic development indirectly drives climate warming, contributing to the increased atmospheric temperatures through e.g. greenhouse gas emissions (IPCC, 2011). Fire management can have a limited effect on the regime shift through fire suppression practices.

Shift from coniferous to deciduous-dominated forest

The coniferous dominated boreal forest of Interior Alaska requires a deep soil organic layer to succeed over other classes of trees. Once the required depth is present, this has a strong reinforcing effect on the coniferous forest. This is a requisite to maintain cold moist soil conditions in which coniferous trees thrive. The cold soil temperature reduces the decomposition rate of organic litter, maintaining a deep soil organic layer (Johnstone et al. 2010).

External shocks such as severe and frequent fires or insect outbreaks might cause dramatic shifts in the soil and vegetation by impacting the soil organic layer depth and tree succession (Hollingsworth et al. 2013; McCullough et al. 1998; Wolken et al. 2011). If the severity of such external shocks is sufficient, thresholds in the most important regime characteristics will be crossed (e.g. shallow post-shock soil organic layer) creating windows of opportunity for deciduous species to dominate. Atmospheric temperature influences the frequency and severity of fires, the soil characteristics and the length of the growing season; therefore, climate warming can also cause a change in conditions that would set the grounds for a shift to a new stable state (Mann et al. 2012).

The deciduous dominated forest has a competitive advantage in warm and dry conditions with shallow soil organic layers. Again, once the required shallow layer is present, it has a strong reinforcing effect on the deciduous forest. A shallow soil organic layer maintains warm and dry soil conditions that speed up the decomposition of organic litter, maintaining the required depth of soil organic layer (Johnstone et al. 2012).

Shift from deciduous-dominated  to coniferous forest 

In order to shift from regime 2 to regime 1, a prolonged climatic cooling would be necessary. A slow change of the environmental conditions towards those that favour a coniferous boreal forest could, if the forest is left undisturbed for a sufficiently long time, lead to late successional coniferous species, like the black spruce, replacing early successional deciduous species (Rupp et al. 2002). Although the spatial scale considered is of extreme importance, sufficiently large external shocks, such as deciduous-specific insect outbreaks, could clear sufficiently large areas to advance the dominance of coniferous trees. In such hypothetical events, coniferous trees could get the opportunity to sprout and start altering the soil conditions to their own favour.

The provisioning of wild products such as berries and game is expected to shift some species and decrease others, but it is unclear what the net changes will be (Chapin et al. 2008). The boreal forest is likely to become a net source of carbon to the atmosphere (McGuire et al. 2009; Schuur et al. 2009), while the air quality is expected to decrease periodically due to the increase in fire frequency (Chapin et al. 2008).

The primary tactic used for reducing impacts of fire and enhance resilience (i.e. maintain structures, feedbacks, functions and processes of regime 1 is fire suppression to reduce the fire severity (Chapin et al. 2008). Decrease in fire severity could prevent the fire from consuming the soil organic layer, which in turn favours the conditions for coniferous tree. Fire suppression should be maintained in Alaska since it minimizes the risk fire imposes to life and property in communities. However, distant fires should be allowed to burn under controlled conditions, since it could have a cooling effect at a regional scale due to the higher albedo of post-fire deciduous species compared to the previous coniferous forest (Chapin et al. 2008).

Reducing air temperature is essential to be able to restore regime 1. Alaska itself accounts for a miniscule proportion of global greenhouse gas emissions and consequently to climate warming (Chapin et al. 2008). Therefore, reducing Alaska’s emissions will not have a major impact on global climate change. Global abatement of emissions is the long-term solution. Regardless of global greenhouse gas emission policies, it is highly likely that recent warming and wildfire frequency will continue for several decades in Alaska due to the multidecadal lag in the climate system (IPCC, 2007). This makes the reduction of emissions an inefficient short-term solution.  Secondary challenges that face efforts to achieve reduced emissions include restricting the economy and changing the focus of developed nations on a continued economic but sustainable growth. Educating global public about the social impacts of climate change in Alaska and other parts of the world (e.g. reduction in ecosystem services) may promote the willingness to reduce emissions (Chapin et al. 2008).

In summary, this means that the regime shift from coniferous to deciduous dominated forest in theory could be reversible through manipulation of e.g. soil conditions or local climate. However, due to projected global socio-economic development and climate change, it is highly unlikely that reversing the regime shift will be possible during the next 100 years.

This regime shift involves an abrupt reorganization of an aquatic food web due to a decrease in top predators or increase in environmental factors like temperature. Trophic cascades play a central role in these regime shifts. During these cascades groups in adjacent trophic levels often show inverse patterns in their abundance (Carpenter 2003). In other words, every other trophic level displays an increase in biomass, which leads to increased predation pressure and biomass decrease on every other trophic level. Most trophic cascades are described in ecosystems with low species diversity and/or simple food webs and/or small size (Frank et al. 2005). The magnitude of a trophic cascade varies depending on species diversity, regional oceanography, local physical disturbance, habitat complexity and fishery practices (Salomon et al. 2010). Food webs with more trophic levels and biodiversity show less trophic cascades than simpler systems (Salomon et al. 2010); due in part to omnivory and the variation on trophic interaction strength (Bascompte et al. 2005). However some documented examples includes subtidal reefs in New Zealand (Shears and Babcock 2002), Caribbean coral reefs (Bellwood et al. 2004), the Gulf of Maine (Steneck et al. 2004), and the North Atlantic (Kirby et al. 2009).

Predator-dominated food web.

Ecosystems in this regime are characterized by high predatory fish, low planktivorous fish, high zooplankton and low phytoplankton abundance. Such ecosystems can be subject to human influence, such as intensive fishery, but exhibit so called “natural compensation” e.g. via species richness and omnivory (Pace et al. 1999) that maintain the regime. Despite high productivity, such ecosystems are less likely to suffer from eutrophication due to high grazing pressure on phytoplankton. Ecosystem services associated with predator-dominated food webs includes food provision, higher biodiversity, better disease and pest control, as well as cultural services like recreation and aesthetic values (e.g. diving, sport fishing).

Planktivore-dominated food web.

Ecosystems in this regime are characterized by low predatory fish, high planktivorous fish, low zooplankton and high phytoplankton abundance. In extreme cases the actual number of trophic levels (TLs) can decrease. Ecosystems with lower trophic level dominance are more vulnerable to eutrophication due to lowered grazing pressure on phytoplankton, as well as invasion of planktivore organisms like jellyfish. A planktivore-dominated food web can affect the fluxes of carbon, transforming aquatic systems -coastal and lakes- from sinks to sources of green house gases (Bakun et al. 2010; Estes et al. 2011). Other ecosystem services like fisheries and recreation are expected to be significantly reduced.

Trophic cascades are usually initiated by fishing pressure that have direct or indirect impacts on the abundance of top predators. The response time to these drivers varies across ecosystems. In the case of coastal areas it has been found that the ecosystem response time to overfishing can vary from decades to centuries where multiple predators exist (Jackson et al. 2001).

Other factors that affect the regime shift are global warming and demand of food and fiber. Global warming have effects on upwelling, water circulatory systems that bring nutrients from the bottom of the sea towards the surface. By altering upwellings, food webs receive either too much or too little nutrients. In both cases the energy transferred towards higher trophic levels diminish as nutrients input variability increase. Demand of food strongly drives fishing, being the main cause of lost of top predators over the world.

Predator dominated food webs occur in environments with low disturbance both from fishing and nutrients input. They usually have four or more trophic levels. The regime is maintained by biotic mechanism that comprise predation and competition both intra and inter species. Abiotic factors also play a role maintaining the regime, by allowing an intermediate level of nutrient inputs from the deep ocean necessary for sustaining the food web.

In general, a combination of low top predator productivity due to poor recruitment or somatic growth combined with high exploitation pressure are conditions under which regime shifts in aquatic food webs are likely to occur (Daskalov et al. 2007). Overfishing can unbalance trophic relations, changing the configuration of competition and predation. Weak competitors may become stronger given less fishing pressure on such stocks, or by favoring the reproduction of it’s preys by fishing its competitor species. In addition, changes in nutrients inputs can represent an advantage for lower trophic levels, allowing them to escape the control of its predators.

As result, the ecological function of certain species groups is changed, allowing over abundance of immediate lower level. For example, the decrease of top-predators allows over abundance of meso predators that in turn reduces planktivorous fish and increase zooplankton. Further increase in nutrients from upwellings can lock the food web in a state dominated by planktivore fish as jellyfish. Low nutrients input, on the other hand, would reduce the amount of energy available for the food web and as a result an abrupt reduction of higher trophic levels is expected.

Shift from predators to planktivore dominated food webs

Commercial fish stocks that are locked into a regime where they are maintained at low levels of abundance has obvious economic consequences on fisheries industry. However, these fisheries related shifts are not always negative, as for example a decrease in top predators can increase the catches of mid-level prey (e.g. shrimps and clupeids) (Frank et al. 2005). When an ecosystem-wide regime shift occurs it affects not only targeted groups, such as commercial fish (Daskalov et al. 2007) but also impacts other ecosystem functions. Biodiversity as well as ecosystem resilience may be reduced by a shift from an ecosystem with high top predator diversity and dominance to a system with lower trophic level dominance. This means that ecosystem may become more vulnerable to both climatic and anthropogenic change. Such ecosystems may more easily suffer from eutrophication, hypoxia and invasion by non-native species. Eutrophication and hypoxia in turn lead to increased algal blooms and decrease the recreational values of water bodies.

Predators and commercial fish stocks have remained under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the reduction of food webs, which in turn increases the risk of fisheries collapse. Depletion of fish stocks has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Hassan et al. 2005). Fish catches are projected to further decrease in the 21st century affecting protein sources for people, especially in poor regions (Hassan et al. 2005). The estimated current contribution of fisheries to human protein consumption is 29 million tons produced industrially and 24 million tons in small-scale fisheries (Hassan et al. 2005).

Options for enhancing resilience

In food webs that exhibit strong top-down control, decreasing the fishing pressure on top predators is an obvious strategy (Scheffer et al. 2001; Frank et al. 2005). However, due to the hysteresis effect, return to a predator-dominated food web may often require substantially lower fishing levels than that which induced the regime shift, together with additional measures. Hence, restoration is usually costly, especially to fisherman (Moellmann et al. 2009). As mentioned above, environmental factors can play an important role in maintaining alternative food web regimes. Nutrient loading is one of the key parameters controlling processes such as eutrophication and hypoxia. However, decreasing nutrient loads in marine environments requires large scale actions and often collaboration at international level (e.g. Helsinki Commission Baltic Sea Action Plan in the Baltic Sea). Hence, the time-frame of these actions in addition to time-lag in ecosystem response is often long. Climatic variation cannot be directly controlled by human actions. However, climate impacts can be mediated by employing other management measures that contribute to maintaining healthy marine ecosystems with high biodiversity and resilience, which then can compensate for changes in climate.

Options for reducing resilience to encourage restoration or transformation

In addition, in small closed food webs systems (3 to 4 trophic levels), biomanipulation has been suggested to manage one of the symptoms of degraded marine environments, namely eutrophication. This involves increasing the population of predatory fish such as bass, pike and walleye through stocking or reduced angling quotas. Increased populations of these predators leads to a decrease in the level of zooplanktivores. This in turn allows an increase in the population of planktivores that graze on the algae and zooplanckton, helping to reduce the algal density and phytoplankton respectively(Smith and Schindler 2009). This option, however, has limited reach when it comes to open marine foodwebs, where seasonal migrations and metapopulation dynamics are common.

During the 20th century global fisheries greatly expanded in terms of effort and range. This expansion resulted in the overfishing of many fish stocks, and in some cases triggered fisheries collapses, for example the collapse of the highly productive Newfoundland cod fishery (Worm et al. 2009). While overfishing can severely deplete fish stocks, it will not necessarily trigger a fisheries collapse. A collapse occurs when once fishery pressure has been removed, stocks fail to rebuild (Hutchings 2000, Kirby et al. 2009), suggesting that the system has become locked into an alternate regime. In this case, alternate regimes are usefully conceptualized as alternative food web structures:

High abundance of a commercial fish species.

In this regime a valuable fish species is common and the fishery is highly productive. The main ecosystem services associated with high abundance of commercial fish are the production of food in first place, employment for both artisanal and industrial fishermen; followed by pest and disease regulation by healthy food webs, recreational values as recreational fishing, knowledge and educational values, as well as maintenance of biodiversity.

Low abundance of the commercial fish species.

This regime is characterized by a substantially reduced abundance of the valuable fish species. In cases where the valuable fish species performs key ecological functions, it can lead to trophic cascades that lead to an overabundance of planktivorous fish or primary producers such as seagrass, macroalgae, sponge or phytoplankton (Jackson et al. 2001). With low fish abundance, food production is the most immediate ecosystem services affected. While industrial fishermen often can move and harvest new areas or new species, artisanal fishermen often can lose their livelihoods and traditions.  

There are two key direct drivers of fisheries collapse: overfishing as a top-down disturbance, and the change in nutrients input as a bottom-up disturbance.

Overfishing reduces the population size of commercial fish species, reduces their ability to perform their functional role in the ecosystem, can destabilize their population dynamics (Anderson et al. 2008), and ultimately could change the structure of the food web. Nutrient inputs, on the other hand, affect fisheries by increasing the abundance of lower trophic levels e.g. phytoplankton and zooplanktivores such as jellyfish (Daskalov et al. 2007). Such changes in food web configuration can reduce the energy transfer to fish and hence reduce fishing productivity. These drivers can reduce the resilience of a regime making it more vulnerable to other types of regime shift, such as climate driven food web reorganization [see: Climate & Marine Food Webs]. The indirect drivers that explain overfishing and changes in nutrient input interact in synergistic ways, making their identification a challenging process (Ocean Studies Board 2006, Kirby et al. 2009).

Indirect drivers of overfishing include oversupply of fishing boats, inflexible fishing quotas, demand for fish, the development of new fishing technologies, the facilitation of trade (market connectivity), governance failures such as the failure of regulation to halt illegal fishing, and market failures such as subsidies or perverse incentives. Nutrient enrichment of marine ecosystems is also indirectly driven by runoff of fertilizer nutrients from agriculture and sewage from urban areas (Diaz and Rosenberg 2008).

A fishery collapses when a high-valued commercial species is substantially depleted and fails to immediately recover when fishing effort is removed (Hutchings 2000). It is usually associated with a substantial and persistent change in the structure of the marine community. Such a change involves changes in the species presence, their relative abundance, size and age.

When the stock of commercial species decreases or even disappears the populations of other similar species may increase. Many commercially fished species are predatory fish, such as tuna, that have a high trophic level. Removing such top predators can cause a trophic cascade to reshuffle a food web (Daskalov et al. 2007, Estes et al. 2011), especially if there are not ecologically similar species present. Such trophic cascades, can increase fish stocks of prey species (Walters and Kitchell 2001), which can in turn lead to either sequential depletion of fish stocks or the development of new fisheries (Essington et al. 2006).

The low abundance of commercial fish species can be maintained by a number of mechanisms. First, when the population of the commercial fish species falls under certain levels their population growth rate can decline from a variety of causes. This is known as depensation in fisheries, or the Allee effect more generally (Liermann and Hilborn 1997). Such an effect can occur if small schools of fish offer poorer protection from predators, smaller populations are less able to hunt or detect food, or mating success rates can decline.

Second, nutrient enrichment can favor the food web shifts towards zooplanktivores, such as jellyfish. By reducing resources for other fish, the rise of such species can suppress fisheries recovery. Fishery pressure on filter-feeding fishes like sardine exacerbates the problem by favoring jellyfish outbreaks (Jackson et al. 2001, Brotz et al. 2012, Condon et al. 2013).

Despite evidence from some well-documented cases, the degree to which fisheries declines are regime shifts remains a controversial issue for a number of reasons (Hilborn 2007, Litzow and Urban 2009). First, time series for most fisheries are only available from 1950, a time when marine ecosystems were already affected by flourishing industrial fishing, and some stocks had already been seriously depleted, especially those of large marine vertebrates like whales, manatees, crocodiles, seals, swordfish, sharks and rays (Kirby et al. 2009). This lack of long term data makes it difficult to identify causation in fisheries time series. Furthermore, marine ecological diversity and food web complexity complicates the detection of changes in fish populations due to fishing, because the signals of regime shift dynamics are small relative to ecological variation (Liermann and Hilborn 1997).

Shift from high to low abundance of commercial fish species

Since 1950 the world marine fisheries catch has expanded almost fivefold, and is currently about 80 million tonnes of fish. According to FAO, about 30% of fish stocks are overexploited, 60% are fully exploited and about 10 % are not fully exploited (FAO 2012). Global marine fish catches have plateaued and are not expected to increase. Local fisheries decline has been estimated to have affected the employment of roughly 14 million fishermen, 12 million of which correspond to artisanal fisheries (Pauly et al. 2006). There is not clear assessment of the number of stocks that have collapsed but FAO suggests about 5% of fish stocks are depleted (FAO 2012). Some collapsed stocks have remained at under 10% of their previous sizes even after decades of fishing closure (Ainley and Blight 2009). Food production, primary productivity, pest regulation, and cultural services are the ecosystem services most affected by the collapse of fisheries.  

Shift from low to high abundance of commercial fish species

An abundant fish population regime will likely better maintain marine biodiversity and will be more resilience to climate variation and climate change. Society will benefit of high abundance regime in terms of high primary productivity that translate in more food, jobs and livelihoods in small coastal communities. It will also better regulate marine hypoxia.

A review of marine fisheries in the Millennium Ecosystem Assessment offer a synthetic set of managerial options for addressing fisheries collapse (Pauly et al. 2006).

Options for enhancing resilience

There are two types of strategies for enhancing resilience. The first is to stop doing things that are eroding resilience, while the second is to being activities that build resilience. There are a number of fishing practices that could be eliminated because they damage the resilience of fisheries. For example, fishing practices that destroy habitat such as bottom trawling should be avoided. Developing alternative technologies and reduction of fishing effort are suggested. Bycatch, the killing but not consuming of unwanted fish, can also have substantial impacts and can be reduced. Indirectly, fisheries subsidies and inflexible quotas can drive fishing effort to high levels that erode the resilience of a fishery.

Options for reducing resilience of unwanted regime to encourage restoration or transformation

Direct resilience building strategies include the establishment of Marine protected areas (MPAs) (Takashina and Mougi 2014). These areas can play an important role by providing refugia for fish and increasing ecological diversity. Research suggests that they need to cover ecologically significant areas, including different habitats, ensuring connectivity among them. However there remains controversy over the extent to which MPAs shift fishing pressure to less protected sites.

Indirect resilience building efforts include efficient regulation and policing of fishing can reduce illegal fishing, while comprehensive fisheries policies at the international and national level can also build resilience if they result in fisheries policies that reinforce one another. Indirect resilience building strategies include improving the ability of fishers to exit fisheries.

Good management of fishing pressure has been able to restore fisheries stocks. For example, in the US fishing closures and implementation of management measures for bottom fishing have enabled the restoration of George Bank haddock, Atlantic scallops, George Bank yellowtail flounder, Atlantic striped bass, Atlantic Arcadian redfish, Pacific chub mackerel, and Pacific sardine. However, there is no clear strategy that works across many ecosystems or nations (Beddington et al. 2007).

Mangroves are ecosystems adapted to the mixture of fresh and salt water common in intertidal zones of marine coastal environments. Mangroves cover between 1.7-1.8 x 105 km² globally. They develop within 30ºN and 38ºS and grow optimally between 15ºC and 25ºC (Lovelock 2008). Outside this geographical and thermal range mangroves show reduced leaf formation and above 38ºC or below freezing temperatures they do not photosynthesize (Mcleod and Salm 2006; Cavanaugh et al. 2014). Mangrove ecosystems are often used by local communities who depend on fishing and timber harvesting.

Mangrove forest

Mangrove forest is characterized by an ensemble of species with aerial roots adapted to survive in environments where water salinity and tides are highly variable. Mangroves grow in tropical and subtropical waters, where the optimal temperature range is between 15-24ºC. Different species occur under different salinity gradients and some trees reach 40m high in areas with high precipitation. Mangroves usually establish in soil enriched by the sediment discharge of rivers (Mcleod and Salm 2006). Mangroves are ecologically important because they provide habitat for many marine species at some stages of their lifespan, increasing both marine and terrestrial biodiversity.

Salt marshes, rocky tidal, shrimp farms.

Mangroves can become dominated by salt marshes or other configuration of coastal wetlands depending on the substrate, sediment deposition and the topography (Mcleod and Salm 2006). The alternative regime once a mangrove has collapsed is not straightforward. It strongly depends on the land use, topological characteristics, latitude and the history of the system disturbance (Cavanaugh et al. 2014). For example, many mangroves around the world have been converted into ponds for shrimp farming, salt extraction, or simply dry out to establish agriculture, cattle ranching, infrastructure development or human settlements.

The main drivers of mangrove collapse have typically been land use/cover change during the last 50 years. One third of world’s mangroves have been lost due to overexploitation of forest resources (deforestation) or conversion into agriculture, salt extraction, or ponds for shrimp farming (Cavanaugh et al. 2014). Other threats to mangroves include the development of infrastructure (roads, dams), the diversion of fresh water for irrigation and the development of urban areas. These impacts are typically stronger in developing countries where it is expected that mangroves will decline 25% by 2025 (Mcleod and Salm 2006), while recent studies show that mangrove areas are declining 1-2% yearly (Duke et al. 2007; Alongi 200; Cavanaugh et al. 2014).

Current climate change is also expected to affect mangroves via the increase in frequency and severity of storms, floods and droughts (climate extreme events), the increase of temperature, sea level rise, and ocean acidification. Frequent extreme events that potentially decrease the flow of fresh water will increase salinity, putting stress on some species and reducing their growing rates and seedling survival (Mcleod and Salm 2006). Precipitation increase, on the other hand, could favor mangrove over salt marshes by decreasing salinity and giving a competitive advantage to mangroves (Cavanaugh et al. 2014). However, strong storms and flooding events could prevent mangroves from respiring when their aerial roots are underwater. Although temperature is expected to increase 2-6ºC before 2100, its effect on mangroves is contested. On the one hand, a 6ºC increase would heterogeneously affect mangroves around the world, although it is likely to induce thermal stress, it won’t be strong enough to surpass the temperature tolerance of these ecosystems. An increase in temperature will increase the melting rate of ice and expand the volume of world’s oceans increasing the level of the sea. On the other hand, it has also been observed that warming on the coldest time of the year can favor mangroves over taking salt marshes in temperate areas of the planet, with a suggested threshold of -4ºC for Florida (Cavanaugh et al. 2014).

Sea level rise is perhaps the most important threat to mangroves. While the expected sea level rise projection range from 0.09 to 0.88 m (or 0.9 to 8.8 mm per year) several studies report that mangrove ability to migrate upwards range from 1.2 to 4.5mm per year (Mcleod and Salm 2006). Although faster migration has been observed in Holocene stratigraphic records for Florida, migration also depends of local topological conditions. Increase in atmospheric carbon dioxide could increase mangroves growth but also ocean acidification. While the increase in growth is not expected to be strong enough to overcome the pressure of sea level rise, acidification is expected to have an indirect effect on mangroves by reducing coral reefs accretion, increasing in turn the erosive effects of waves on mangrove soils (Mcleod and Salm 2006).

Shift from mangroves to salt marshes, rocky tidal or shrimp ponds.

By having aerial roots mangroves adapt to very specific conditions where salt and fresh water meets in coastal zones. Mangroves build up carbon rich soils also known as peat, which favors further mangrove growth.

Loss of mangrove area decreases peat production. Peat is lost by wave erosion and reduces the likelihood of recovery of mangrove forest. Mangrove area loss is mainly driven by conversion to agriculture, urban settlements, overexploitation of wood, shrimp farming, salt extraction, or infrastructure development (e.g. roads, dams, water divergence to irrigation) (Cavanaugh et al. 2014). Human perturbations are maintained by social feedbacks such as demand for food and market access, population growth and the cumulative benefit of human settlements (higher provision of services). In contrast, communities with poor services access such as electric energy or gas for cooking strongly rely on mangrove wood as source of energy and construction material (Mcleod and Salm 2006).

If sea level rise occurs faster than the ability of mangroves to build up peat or migrate to higher areas, mangrove forests will disappear in major areas of the world (Alongi 2008). Both the impact of sea level rise and mangrove’s ability to adapt depend on local conditions and are highly uncertain. For example, mangroves in carbonate substrates (such as coralline islands) and microtidal systems (short tidal change from high to low) are less likely to adapt and migrate as climate change continues. Mangroves in rich soil substrates and macrotidal systems are more likely to migrate if salinity balance and sedimentation do not overcome their tolerance range (Mcleod and Salm 2006).  

Shift from salt marshes to mangroves  

Mangroves have also been observed to migrate pole wards colonizing habitats previously dominated by salt marshes (Cavanaugh et al. 2014). The shift occurs when the minimum temperature in winter increase for a longer period of time, giving mangroves an advantage to over compete salt marshes. It has been observed in areas such as Florida, Louisiana or Australia. Although there is uncertainty regarding the mechanism, the threshold is thought to be related with the number of days that minimum temperatures are colder than -4ºC, at least for Florida case (Cavanaugh et al. 2014). The authors of the Florida study also report that it’s unlikely that local scale drivers such as nutrients inputs, sedimentation or hydrology have had a determinant effect on the shift. However, sea level rise has contributed to increasing mangrove area (since it has happen slowly enough) and other common driver of the shift can reduce mangrove accretion such as coastal development, aquaculture and timber production (Cavanaugh et al. 2014).

Mangroves provide a variety of ecosystem services. Mangrove ecosystems provide habitat for economically important marine species for fishing (e.g. shrimp or lobster among other crustaceans and mollusc), and they provide habitats for important groups of terrestrial species such as reptiles, monkeys, and birds. They are therefore important for maintaining both marine and terrestrial biodiversity. Mangroves also provide fuelwood, charcoal, and wood for construction. As a forest it provides protection against storms, floods, river born siltation and also traps water pollutants and sediments, reducing the erosive action of waves, particularly protecting adjacent ecosystems such as coral reefs, and sea grass beds (Duke et al. 2007). Mangroves have also been shown to play a key role in the global carbon budget, capturing up to 15% of global carbon and exporting up to 11% of particular terrestrial carbon to the oceans (Alongi 2014). In fact, recent studies support the hypothesis that mangroves are able to capture and store carbon in a greater extend that terrestrial ecosystems (Lovelock 2008), nearly identical to those of tropical forest (Alongi 2014). Therefore, losing mangrove area implies the loss of carbon storage.

While in the short term local communities could benefit by converting mangroves into shrimp farms, agriculture fields, or using its wood; in the long term they mean the loss of important services such as coastline protection, biodiversity important for tourism and fishing, water cleansing and carbon storage (Ewel et al. 1998; Cavanaugh et al. 2014). In fact, their services has been valued over $1.6 trillion per year (Costanza et al. 1997).

McLeod et al. (2006) propose a series of tools to monitor and plan for mangroves adaptation to climate change. First they propose to assess mangrove vulnerability based on the local conditions. Management options for mangrove forests are highly context dependent. As the main threat to mangroves is climate change through sea level rise, mangrove ecosystems will likely migrate upwards and northwards (Cavanaugh et al. 2014). However, such migration is constrained by local conditions such as substrate types, infrastructure development, sediment input and changes in salinity.

Second, they propose to apply risk-spreading principles (Duke et al. 2007) and identify areas that are likely to be sources and sinks of the migratory process (Mcleod and Salm 2006).  By identifying the role of each mangrove patch, it is easier to prioritize where green belts are needed to increase connectivity, which patches should be protected and which patches are more likely to respond to restoration efforts. They emphasize the importance of establishing a baseline and monitoring program with local communities in order to assess the main drivers locally. Developing partnerships with local stakeholders and developing alternative livelihoods for mangrove dependent communities are key managerial efforts (Mcleod and Salm 2006). The resilience of mangroves to future climate change scenarios also depend on how successful management efforts are at reducing the influence of other drivers: deforestation, overexploitation, infrastructure development, divergence of water for irrigation, and urbanization. Management of such local forcing will increase resilience to regional forcing such as increasing storm events or ocean acidification.

Third, it has been suggested that policy and market-based are critical to foster conservation efforts, reduce mangrove encroachment and maintain carbon storage through funding mechanisms such as REDD or other payment for ecosystem services schemes (Hutchison et al. 2013). Effective governance structures and education has also been suggested as managerial strategies (Duke et al. 2007).

Co-operation and sustainable resource levels

Harvesters extract from the resource at rates that are socially optimal, ensuring that resource levels stay at their most productive level and the community-average payoff is high. Social capital exists within the community, ensuring any defectors are ostracised.

Over-harvesting

Harvesters exert high efforts in extracting the resource, leading to depletion of the resource and low payoffs for the community. Social capital is absent and defectors are not ostracised.

Shift from ‘Co-operation and sustainable resource levels’ to ‘Over-harvesting’

The regime shift occurs when ostracism ceases to be an effective mechanism for encouraging defectors to co-operate, because the benefits of overharvesting begin to outweigh the disadvantages of ostracism. A number of factors could drive this shift. Increasing resource level, for example due to increased inflow, can lead to ineffective ostracism because at higher resource levels defection becomes more attractive by providing higher gains from resource over-extraction. Increasing defector payoff compared to ostracism, for example due to decreased costs or decreased ostracism strength, could also lead to defection becoming increasingly attractive.

Shift from ‘Over-harvesting’ to ‘Co-operation and sustainable resource levels’

Reversal of any of the above trends can cause a shift from over-harvesting to co-operation: decrease in resource levels, for example due to decreased inflow; increased costs; or increased ostracism strength.

Co-operation and sustainable resource levels occur when there is sufficient social capital to ostracise defectors and thereby make defection less attractive than co-operation. The key negative feedback that keeps the regime stable is the following: if the fraction of cooperators increases the social capital and hence strength of ostracism increases which reduces the utility of defectors, making defection less attractive. At the same time, however, resource productivity increases which increases the utility of defectors, so stability ultimately depends on the respective strength of the two feedback loops.

The over-harvesting regime exists when there are only few cooperators and hence there is insufficient social capital to ostracise defectors. The feedback that maintains full defection is as follows: any agent that 'co-operates' (i.e. harvests at a lower level) will have a payoff substantially less than the defectors. In the absence of social capital to encourage co-operation through ostracism, the co-operator will switch back to defection. The key threshold for a collapse in co-operation is when the costs of ostracism are lower than the benefits of overharvesting, i.e. ostracism ceases to be an effective mechanism to discourage defection. A variety of drivers can cause this shift, as discussed below.

To trigger a transition from full defection back to co-operation, co-operation must become more attractive than defection. This could be achieved by increasing the costs of harvesting, or by decreasing the level of the resource to low levels so that the benefits of overharvesting are minimal. However, once the co-operation strategy is lost in the community, it may be very difficult for it to re-emerge.

The ability of the natural system to provide resources for harvesting is lost with the regime shift from co-operation to over-harvesting (gained for over-harvesting to co-operation). Depending on context, loss of other ecosystem services many accompany the decline in resources. The income that members of the harvesting community obtain by harvesting is severely decreased by this regime shift. Income may even no longer exceed the costs of harvesting and the community may need to find other means of survival.

Options for preventing regime shift to over-harvesting

Management actions that stop the drivers discussed above reaching their thresholds may help to prevent regime shifts. Activities that strengthen social norms and trust in the community and thus enhance cooperation and decrease the incentive to defect and overharvest for the individual benefit (hence increasing the strength of the ostracism).

Options for restoration of co-operation:

Introducing measures that build trust in the community and disincentivise free riding. 

Oligotrophic regime

Coastal regions serve as an attraction to people, and contribute to marine resources, recreation and transport, which on the other hand cause high human impact on these ecosystems. An oligotrophic coastal system has clear water and often submerged vegetation. Commercially preferred fish species may be abundant, because coastal ecosystems support many fisheries and are highly productive due to the nutrients from the land via runoff, or deep ocean via upwelling (Boesch 2002). Many seas experience some natural hypoxia/anoxia and eutrophication, but in smaller extent and shorter time period than in the eutrophic regime. The limited amount of nutrients keeps the growth of primary production and algal blooms restricted.

Eutrophic regime

Estuaries and coastal marine ecosystem receive anthropogenic pollutants rapidly from rivers and streams from drainage basins. In eutrophic marine ecosystems phytoplankton and epiphytic algae biomass, as well as nuisance blooms of gelatinous zooplankton, have increased. Changes in species composition can take place in all trophic levels. Phytoplankton species may have shifted to taxa that are toxic or inedible. Animal species composition often changes to include less recreationally and commercially desired species. Macroalgal and vascular plants can experience changes in biomass and species composition or disappear completely. (Smith et al. 1999; Smith 2003)

Eutrophic regime causes economic losses due to restoration and damaged goods and health threats to humans exposed to algal toxins. Water clarity is reduced and humans perceive the water overall as less aesthetic.

Eutrophication commonly takes place in marine coastal waters (Smith et al. 1999; Cloern 2001). The primary cause of marine eutrophication is excessive increase in nutrient concentration (Nixon 1995) from riverine loads, originating from fertilized agricultural areas, urban sewage and industrial wastewaters (e.g. Bonsdorff et al. 1997). The nutrient loads from land to sea have been successively increased as a result of land clearing, population growth, industrial development, increased use of fossil fuels and increased use of fertilizers in agriculture (Boesch 2002). In addition, oceanic upwelling transports nutrient-rich waters to water surface and atmospheric nitrogen can enter the ocean (Boesch 2002; Paerl 1997). Estuaries may be naturally eutrophic as the nutrient loads from land are concentrated in confined channels, but cultural eutrophication often increases the nutrients significantly. In marine systems nitrogen is usually the key limiting nutrient (Borysova et al. 2005), but the limiting nutrient might differ between locations.

Also other human impacts contribute to the changes observed in many coastal areas (Cloern 2001), such as fishing (removal of top-predators causing food web reorganizations). For instance in the Black Sea the intense eutrophication is suggested to result from the combination of nutrient inputs, low grazing pressure on phytoplankton and favorable climatic influences (Llope et al. 2011). Climate change is projected to intensify eutrophication by potentially increasing the water temperature and thereby strengthening the vertical stratification potentially leading to larger deep anoxic water volumes or by increasing the discharge due to higher rainfalls (Justic et al. 2009). Local characteristics of the sea, such as water residence time, stratification and tides affect the intensity of eutrophication and the ecosystem vulnerability to it (Cloern 2001; Justic et al. 2009).

The biological growth and algal blooms in an oligotrophic sea are restricted by the limited availability of nutrients, in particular nitrogen and phosphorus. Diverse processes tend to regulate algal growth in coastal environments, for instance herbivores control algal biomass and elevated oxygen concentrations suppress recycling of nitrogen and phosphorus (Kemp et al. 2005).

In summary, nutrient addition increases algal biomass and after substantial nutrient loading affects bottom-water oxygen and nutrient recycling, water clarity and benthic primary production (Kemp et al. 2005). Biomass increases lead to a high remineralisation of organic matter at the sea bottom leading to a drecrease in oxygen concentration (Borysova et al. 2005). Oxygen within sulphates is used by bacteria, resulting in the release of sulfur that will capture the free oxygen still available in the upper layers. Sulfur also causes the specific smell of sediments in eutrophic water. As the deep water oxygen concentration decrease towards anoxic conditions, benthic orgranisms will decrease and eventually disappear.

Positive feedback loops are found in nutrient recycling and water clarity. In the nutrient feedback loop, increase in nitrogen and phosphorus increase the amount of algae and turbidity, which decreases oxygen concentration in deep water, leading to lower redox and denitrification processes. This causes increased recycling of nutrients (ammonium, phosphate), and increases the nutrient availability causing a higher phytoplankton biomass. In the coastal water clarity feedback the increased turbidity and abundant phytoplankton leading to less light and benthic production, which results in less nitrogen and phosphorus uptake and increased resuspension, which inhibit the growth of seagrass once again means more algae and turbidity. (Kemp et al. 2005, Nyström et al. 2012).

In eutrophic systems nutrient availability regulates primary production levels. The increased occurrence of harmful algal blooms are of great concern in marine systems (Smith et al. 1999). Provisioning ecosystem services are lost when toxins cause mortality of both wild and farmed fish as well as shellfish, and other animals. The composition and structure of food webs may be altered (Caddy 1993; Smith et al. 2006), potentially affecting the resilience of the coastal ecosystem. Damage and destruction of habitats (e.g. Kemp et al. 2005; Walker et al. 2001; Andersen & Rydberg 1988) and weakened quality of nursery and spawning grounds (Borysova et al. 2005) cause commercial fish species to migrate away or die.

Over half of the world population lives within coastal areas and most of the world's fisheries are connected to estuaries and near-shore habitats (Smith 2003 ref. Hobbie 2000), making the marine eutrophication a serious problem. Humans experience illness or even mortality due to contaminated fish and seafood and harmful algal blooms. Economic losses are caused by water quality problems, expenditures undertaken to reduce or avoid the damaging effects, damage on market goods and decreased tourism (Committee on the Causes and Management of Eutrophication, Ocean Studies Board, Water Science and Technology Board 2000; Cloern 2001).

Options for preventing the regime shift

Returning to the pristine, oligotrophic state is seldom a realistic option especially with continuation of anthropogenic nutrient inputs (Boesch 2002) and the human population growth. Eutrophication can be avoided, or restored, by defining a good ecological status for example in the European coasts through the implemented Water Framework Directive (WFD), for the system and the standards (nutrient reductions) needed to achieve it (Boesch 2002). The effect of current regulations to meet these standards is estimated and additional means to satisfy the established ecological goals identified, as was also done by the first assessment of the success of the WFD. Overall, the management strategy should be integrated to include for instance the catchment basin and atmospheric impact. The use of nutrients can be reduced already in the point of release and pollution should be removed before they end off at the sea. Ecosystem based fisheries management could improve the resilience of food-web against perturbations (Llope et al. 2011)

Options for restoration of desirable regimes

The key action to decrease eutrophication is limiting (Smith 2003; Smith et al. 2006; Conley et al. 2009) and most measures needs to be taken in the surrounding catchments. Commitments and organized efforts have been made in several regions to reverse eutrophication. Examples of institutional arrangements are regulations to reduce emissions, multinational directives for the national level such as the WFD, coastal management programs and educational efforts (Boesch 2002).

Different ecosystems need separate, specific management frameworks including several sectors. Local to regional nutrient reduction programs are formed in combination of scientific knowledge and politics. Reductions can be done through wastewater treatment, emission regulations, increased efficiency of nutrient use in agriculture, reduced fertilizer use, manure management, enhancement of nutrient sinks and reduction of urban runoff (Boesch 2002).

Estimation of progress requires observations, continuous monitoring of the nutrient sources and tracking the changes in the sea back to the particular action. A time lag of years may occur before the management effects can be seen in the sea, in particular as internal phosphorus loading from sediments continues.  Modeling enables forecasting the future consequences for nutrient reduction for, for instance, water clarity, algal blooms and oxygen levels. (Boesch 2002)

Intact WAIS (full glacial or modern interglacial)

Antarctica is divided into two large ice sheets by the Transantarctic Mountains. A large part of Antarctic ice mass consists of the WAIS, which lies on a marine sediment-draped bed many hundreds of meters below sea level (Joughin & Alley 2011; Bamber et al. 2009). Compared to the East Antarctic ice sheet, the WAIS has rapid ice flow and discharge (Anderson et al. 2002). Ice shelves, which form over the water and float permanently attached to the landmass, serve as buttresses for the inland ice streams of the ice sheets (permanent layer of ice covering land), and are vulnerable to environmental forcing. Ice volume is stable when the water entering and leaving the system are in balance; the net accumulation (mainly snow) balances losses (ice calving, basal melting at the ice sheet margins) (Ivins 2009). The stability of the marine ice sheet is created between local basal restraint of the inland ice and longitudinal stretching of the ice shelves (Joughin & Alley 2011; Ivins 2009), meaning that the marine-based bottom topography, geometry of the interior part and stability of the buttressing ice shelves (Joughin & Alley 2011) contribute to maintaining the regime of the intact WAIS. The Southern Ocean marine environment generally rich in diversity, whereas the Antarctic terrestrial environment is low in diversity and missing many taxonomic groups (Rogers et al. 2012). In this regime humans have harvested large marine mammals, fish and krill, the latter of which is in direct competition with higher trophic level species (summer breeding colonies) (Rogers et al. 2012). At the present the WAIS appears to be thinning and the extent of sea ice is increasing.

Disintegrated WAIS (extreme interglacial)

Changed internal dynamics, such as complicated ice stream flow changes or natural climatic and oceanic forcing, create instability in the WAIS. Disintegration (collapse) of the WAIS means the nonlinear process, in which the ice sheet would flow at increasing rates into the ocean as its buttressing ice shelves diminish (O'Reilly & ACOS 2013). In the absence of the WAIS, the area would be covered by a broad open sea, punctuated by islands (Scherer 1991). The changes are occurring now and the collapse of the WAIS appears to already be underway, but large uncertainty remains on the timing (Joughin et al. 2014; Joughin & Alley 2011). The absence of the WAIS would affect local ecosystems adapted to the presence of ice, and due to the sea level rise, all global coastal ecosystems and coastal human habitats. Marine fauna adapted to the sea ice dynamics would be directly impacted through habitat changes and shifts in marine isotherms, and the changes could cascade to the higher trophic levels, and alter food webs (Rogers et al. 2012; Clarke et al. 2007). The absence of WAIS would leave broad, deep seaways (Joughin & Alley 2011), increasing accessibility for fishing vessels and tourists.

Climatic and oceanic forces appear to cause the WAIS retreats. Geologic evidence shows that the Antarctic ice sheet has shrunk in the past when global temperatures were warmer, for instance during the late Quaternary period (see e.g. Barnes & Hillenbrand 2010). There have been times when parts of the WAIS have been lost to the ocean, causing raised sea levels (Ivins 2009). According to the model by Pollard and DeConto (2009), a collapse from the modern conditions could occur when sub-ice ocean melting increase from 0.1 to 2 m yr^-1 under shelf interiors, and from 5 to 10 m yr^-1 near exposed shelf edges. The mechanisms leading to the WAIS collapse are still unclear. It has been suggested that the WAIS collapse could start by the intrusion of ocean water (i.e. sea level rise) between the ice sheet and the ground. As the grounding line retreats, a strong positive feedback would be triggered when ocean water undercuts the ice sheet and causes further separation from the bedrock (Oppenheimer 1998). This feedback with sea level changes driving further retreat is contested by other studies (Alley et al. 2007; Gomez et al. 2010). At present, main cause for the thinning of the WAIS ice mass is suggested to be the intrusion of relatively warm ocean water beneath the ice. Ice shelves are very sensitive to ocean temperature changes (Turner & Overland 2009) and several studies (see e.g. Price et al. 2008; Thoma et al. 2008, Joughin et al. 2014) have shown that increased transport of warm subsurface water significantly contributes to the basal ice-shelf melt. The warm Circumpolar Deep Water (CDW) protrudes into the ice shelves through submarine troughs (Bintanja et al. 2013) and produces large melt rates at the regions where it is able to access the sub-ice-shelf cavities (Jacobs et al. 1992). Although the upper layers of Southern Ocean have cooled, the subsurface sea has in fact warmed (Robertson et al. 2002). The reasons for the warming of the Southern Ocean are not completely know. It is probably caused by multiple processes, such as increased greenhouse gases in the atmosphere, shifts in Southern Annular Mode (SAM, a high-latitude mode, also called Antarctic Oscillation) and changes in sea currents.

Other, less direct contributors to the regime shift found in literature are hydrofracturing of crevasses by surface melting water, changes in landscape (glacier behavior), wind forcing, ocean upwelling, tides and long period waves, changes in stratospheric ozone affecting atmospheric and ocean circulation, and intensification and southward migration of the Southern Ocean Westerlies due to the positive trend in the SAM (e.g. Thompson & Solomon 2002; Goosse et al. 2008; Turner et al. 2009; Joughin et al. 2014).

In the intact WAIS regime there is variability in the system depending on the natural climate variation, but the rate and scale of the changes in ice mass are not as remarkable as in the second regime, and the ice volume is maintained by equal net balance. The WAIS icescape is at the present characterized by decrease in ice mass and increased sea ice extent. The most recent research indicates that the WAIS is changing significantly and rapidly, and at an accelerating rate (O'Reilly & ACOS 2013; Abram et al. 2013; Joughin & Alley 2011; Thomas et al. 2004; Velicogna & Wahr 2006). West Antarctica is currently warming, and has been stated to be one of the most rapidly warming regions on Earth (Bromwich et al. 2012; Steig et al. 2013).

The possibility for the collapse of the WAIS has been debated in science since the 1970s (Notz 2009). It has been suggested that the WAIS retreats follow glacial and interglacial periods (Clark et al. 2002; Fairbanks 1989; Scherer 1991). The potential of the WAIS to collapse and the mechanisms leading and preventing it are still unclear, but the recent studies argue that the collapse is possible with small changes in the forces (Mercer 1978; Oppenheimer 1998; Vaughan 2008; Schoof 2007; Naish et al. 2009; Pollard & DeConto 2009; Notz 2009). A basin-scale ice-flow model by Joughin et al. (2014) provided strong evidence that the early-stage collapse has already begun. Their model strengthens the argument that the losses are melt-driven and that melt-induced ice-shelf thinning reduces buttressing, creating far greater speedup and retreat through the grounding line retreat. Ice sheet thinning appears to take place through basal melting (i.e. on the underside of the ice shelves) (Pritchard et al. 2012; Joughin & Alley 2011). The highest melting rates occur where the ice shelves interact with the warmest water. Increased atmospheric temperature and ocean temperature (possibly due to several factors such as global climate change, natural climate variation, ozone hole and changes in atmospheric and oceanic circulation) cause intrusion of warmed ocean water beneath the ice shelf of the WAIS. Relatively warm Circumpolar Deep Water (CDW) protrudes into the ice shelves through submarine troughs and produces large melt rates at the ice/ocean interface, decreasing the volume of the ice mass. The estimations of the loss rate vary both regionally and for the whole continent, see for instance King et al. (2012). Although in some regions the ice sheet is thickening, the net balance is negative (e.g. Velicogna & Wahr 2006).

The mass loss of ice is expected to increase both in volume and rate in the future when the warm circumpolar deep water is able to reach further ice shelves (Rignot et al. 2011; Hellmer et al. 2012). Unlike in the Arctic and as an unusual feature in consideration to the global warming, the extent of the Antarctic sea ice has increased in the observed period starting from 1979 (Turner & Overland 2009). Sea-ice expansion may be meltwater-induced (Bintanja et al. 2013). Meltwater from the ice shelves has lower density and thus accumulates in the top ocean layer (Price et al. 2008). This upper layer water gets fresher, and resulting cold halocline reduces the convective mixing, causing the atmosphere to cool and freeze the upper 100 m more easily (Bintanja et al. 2013). The subsurface ocean warming, mass loss of ice due to the basal melt and expanding sea ice may constitute a negative feedback loop (Bintanja et al. 2013; Zhang 2007). The fresh melt water has low density and thus accumulates in the top layer, stabilizing the ocean and resulting in less mixing between cold and warm water (Bintanja et al. 2013). The cold and fresh upper seawater layer is easier to cool by the atmosphere and freezes. This meltwater-induced sea ice expansion hinders the warm deepwater from mixing with the surface water. Another feedback, which is positive but at present contested, has been suggested to be formed by the raised sea level further undercutting the ice sheet and triggering its separation from the bedrock.

Humans mainly use the Antarctica for commercial fishing, conservation, research, and tourism. The Antarctic region also plays a remarkable role in the planetary functions. Shifting the regime to that of the fragmented WAIS would affect the regulating ecosystem services of the Antarctica, such as global climate and ocean circulation through ice and snow albedo, CO₂ uptake and icescape changes. The Southern Ocean has a rich biodiversity, and fisheries include krill and toothfish. The biogeochemical cycles of the Southern Ocean and the sea ice impact the structure and dynamics of the marine ecosystems (e.g. life cycles of organisms), especially the trophic levels that are adapted to the presence, seasonality and properties of ice (Massom & Stammerjohn 2010; Arrigo 2002; Brierley & Thomas 2002; Thomas et al. 2010; Eicken et al. 1995; Moline et al. 2008; Tynan 2010; Clarke et al. 2007; Rogers et al. 2012). Sea level rise may have severe impacts on coastal areas by submergence or increasing flooding and erosion, changing ecosystems, and increasing salinization (Nicholls et al. 2011). Fisheries management and accessibility would change due to the altered sea ice extent (Anon 2009).

The absence of the WAIS would leave broad, deep seaways (Joughin & Alley 2011) and result in forced displacement of coastal population and economy (Nicholls et al. 2011). The economic consequences depend on the time-scale: playing out over a century. The WAIS collapse would damage many coastal communities, whereas occurring over several centruries, the additional time would give possibilities to develop an appropriate risk-mitigation strategy (Dowdeswell et al. 2008; Mercer 1978; Joughin & Alley 2011). Increased vessel access to more southern locations may have implications on the safety of ocean navigation (Anon 2009). Tourism, the largest commercial activity in the Antarctic, is predicted to increase due to the extended season with reduced sea ice thickness (Anon 2009). The increased human presence might be a threat to ecosystems, for instance winter ice has usually provided a relief from the fishing pressure.

Two new studies in 2014 state that since the WAIS has reached the early stage collapse phase, the shift is unstoppable (Joughin et al. 2014, NASA in press). Joughin et al. (2014) argue that it is difficult to foresee stabilization of the system unless CDW recedes sufficienty to reduce present level of melting.

Seagrass beds are marine ecosystems that can be found in the subtidal and intertidal zones in the majority of oceans worldwide (Orth et al. 2006). Regime shifts in these systems have been identified as a transition to either an algae dominated regime (Valentine and Duffy 2006; Burkholder et al. 2007; Nyström et al. 2012) or a barren sediment regime (van der Heide et al. 2007, 2011; Nyström et al. 2012), although most literature interchangeably describes these phenomena as seagrass decline and not regime shifts. Although this review uses a social-ecological systems lens, the focus lies on the shift in the ecosystem, and humans are considered as beneficiaries of ecosystem services and as drivers of system change.

Seagrass dominated regime

Healthy seagrasses form large beds, usually dominated by one seagrass species. They are considered ecosystem engineers as they significantly modify the abiotic conditions of their ecosystem to benefit their own success, by reducing current speed, stabilising sediments and creating oligotrophic conditions by trapping inorganic and organic material (Duarte 2002; Orth et al. 2006; Burkholder et al. 2007). Together with epiphytic algae seagrasses form the basis of complex food webs, making these systems highly productive (Valentine and Duffy 2006). Seagrass beds also support high biodiversity and provide important habitats, refuges and nursery grounds for a variety of species, many of which are commercially and ecologically important (Orth et al. 2006).

Algae dominated regime

This regime is characterised by dominance of macroalgae (algae attached to the bottom sediments that can form extensive beds), phytoplankton (free-living, planktonic algae) or epiphytic algae (algae growing on the surface of seagrass leaves), or a combination thereof (Cardoso et al. 2004; Burkholder et al. 2007). They are inherently superior competitors to seagrass, particularly in high nutrient conditions such as eutrophication (Valentine and Duffy 2006), which is a characteristic of this regime. Under such conditions seagrasses are prone to smothering by epiphytes and encroachment by opportunistic macroalgae that can form beds that are resistant to seagrass recolonisation (Valentine and Duffy 2006). Shallow waters tend to be dominated by macroalgae and epiphytes, while deeper areas are dominated by phytoplankton (Burkholder et al. 2007). In general this regime supports lower biodiversity as the variety of habitats associated with seagrass beds are not provided (Duarte et al. 2006).

Bare sediment regime

A third alternate regime is that of a barren sediment landscape (van der Heide et al. 2007, 2011; Nyström et al. 2012). The shift can occur due to extensive removal of seagrasses or sudden disease outbreaks causing large seagrass die-off (van der Heide et al. 2007, 2011). In this regime sediments can easily be re-suspended, causing high turbidity and light attenuation due to a reduction in, or depletion of, the seagrass engineering function (van der Heide et al. 2011). The benthic sediments are coarse in comparison to the seagrass regime and can seasonally host macroalgal beds. Biodiversity is low and the community structure different compared to seagrass beds, as the seagrass habitats are removed (Cardoso et al. 2004).

Shift from Seagrass Beds to Algae Dominated Regime

The shift from seagrass beds to algal dominated state is driven by multiple stressors, but nutrient loading and overfishing stand out as key drivers (Burkholder et al. 2007). Seagrasses are dependent on high influx of light, oligotrophic conditions and sediment bottoms (Eklöf 2008). In coastal waters eutrophication causes an increase of epiphytic or macroalgal biomass, while in shallow parts it will cause phytoplankton blooms, reducing light penetration to a level that will no longer sustain seagrasses but promote algal dominance (Orth et al. 2006; Burkholder et al. 2007). The drivers of eutrophication are anthropogenic, such as nutrient input from agriculture, aquaculture and sewage (Duarte 2002; Burkholder et al. 2011).

There is evidence that altering food webs through overfishing has similar effects, or can further augment the effects of eutrophication, by reducing herbivory, thus releasing algae from the pressure of grazing which can lead to a shift in regimes (Heck Jr and Valentine 2007). Changes in food webs are in turn linked to increased coastal migration, tourism, increased unemployment rates and increased population in coastal areas (Eklöf 2008).

Another important driver is the sediment load in the water column since it contributes to turbidity, i.e. decreases light penetration. Erosion, coastal development and deforestation are the main drivers of increased sediment loading in the water column (Duarte 2002; de Boer 2007). Activities such as boating, anchoring, dredging and trawling can also affect seagrass beds negatively throughout a long period of time. They all cause water turbidity and resuspension of sediments as well as physical damage to the seagrasses. Resuspension of sediments can also lead to a release of nutrients, promoting algal growth (de Boer 2007).

Shift from Seagrass Beds to Barren Sediment Regime

Another shift that can occur is from seagrass beds to a barren sediment regime with increased turbidity and where seasonal macroalgae take over (Cardoso 2003). Drivers of this shift are mainly physical disturbances such as actual removal of beds from e.g. beach replenishment or dredging, and a wasting disease that can cause extensive seagrass die-off (Duarte 2002; Cardoso et al. 2004). These drivers are shocks to the system, thus it appears that the shift is more abrupt than the shift to the algae dominated state. Physical disturbance could either be anthropogenic, such as dredging, boating activities, trawling and various coastal developments; or natural, such as storms (Duarte 2002). Eutrophication can also be a driver for this shift, as in the Mondego estuary in Portugal, where the loss of seagrasses and their ability to bind sediments also resulted in a bare, coarse sediment regime (Cardoso et al. 2004).

The seagrass regime and associated feedback loops

Seagrass beds occur under low nutrient, clear water conditions created by seagrasses modifying the abiotic environment to favour their own growth (Burkholder et al. 2007; de Boer 2007). Herbivory keeps algal abundance low and facilitates healthy seagrass beds. Feedbacks maintaining the seagrass bed regime are primarily seagrasses reducing turbidity through absorption of nutrients and sediment stabilisation. Thus, light conditions are improved which otherwise work as a limiting factor for seagrass growth (Burkholder et al. 2007). Additionally, the sediment stabilisation reduces the resuspension of nutrients in the water column, which further controls algal blooms and prevents eutrophication (Duarte 2002). These feedback loops maintain the seagrass dominated regime.

External drivers can slowly undermine seagrass resilience by reducing the engineering function which hampers conditions for seagrass growth. This can eventually lead to a sudden shift to algal dominance due to a reversal of the aforementioned feedback loops once the thresholds have been crossed (Cardoso et al. 2004; Nyström et al. 2012). Such shifts show hysteretic behaviour. That is, a return to previous seagrass regime is difficult due to the alteration of feedback loops (Nyström et al. 2012) and the state can usually not be recovered by re-establishing previous environmental conditions (Scheffer et al. 2005).

Shift from Seagrass Beds to Algae Dominated Regime

The key drivers causing seagrasses to shift to an algal dominated regime are nutrient loading and overfishing (Valentine and Duffy 2006; Burkholder et al. 2007). Nutrient loading from agricultural run-off and sewage, will increase the available nutrients in the water column and possibly lead to a eutrophied state. Algae are superior competitors to seagrass under high nutrient levels and nutrient enrichment will therefore promote algal blooms (Duarte 2002; Valentine and Duffy 2006). This in turn decreases light conditions through an increase in turbidity by phytoplankton, shading by macroalgae and overgrowth (fouling) by epiphytic algae (Burkholder et al. 2007). Moreover, the phytoplankton blooms are further fuelled by decomposing algae and seagrasses as this releases nutrients into the water column (Eklöf 2008). It is difficult to determine a threshold of critical nutrient concentration since it is case specific and depends on factors such as current velocity and herbivory (Valentine and Duffy 2006; Burkholder et al. 2007). A more specific threshold is light availability and if the conditions drop below seagrass tolerance it can cause the regime to shift (de Boer 2007); however, light tolerance is species-specific (Orth et al. 2006).

Overfishing modifies community structure causing trophic cascades due to removal of top predators. This causes an increase in meso-predators which in turn will reduce herbivores through higher levels of predation. A reduction in herbivores releases epiphytic algae from grazing pressure, which leads to algal overgrowth on seagrasses. This eventually suffocates them by obstructing light, and since seagrasses engineer their own habitat the loss will make the environment less suitable for a recolonisation, due to resuspension of sediments (Heck Jr and Valentine 2006; Duarte 2002). Seagrass beds are dependent on grazers to feed on algae before they get too thick to be eaten (Heck Jr and Valentine 2006). The threshold in amount of herbivory is hard to identify but it can be reduced to the same threshold associated with light conditions (de Boer 2007). Herbivory is important as it enhances the resilience of seagrass beds to nutrient enrichment by keeping algal population in low abundance (Valentine and Duffy 2006).

Once established, the algae dominated regime is maintained by the loss of seagrass engineering capacity which keeps the system in a high turbidity and nutrient state. Once macroalgal beds are established they are highly resistant to seagrass encroachment since they are better competitors under high nutrients levels (Valentine and Duffy 2006). In this regime oxygen depletion (anoxia) in the sediments is common due to higher levels of decomposing organic material and algal respiration as these processes consume oxygen, which eventually can result in hypoxia. This further prevents seagrass recolonisation as it causes an increase in hydrogen sulphide concentration in the sediments which is directly toxic to seagrasses. Such toxicity can also result in a decline in herbivores, thus reduce herbivore pressure (Burkholder et al. 2007) and further reinforce the algal dominance.

Shift from Seagrass Beds to Barren Sediment Regime

Both physical disturbance and diseases can remove extensive seagrass cover in one single event, which could eventually lead to a sudden collapse of the whole population and cause the regime shift (van der Heide et al. 2007). Although not specified, van der Heide et al. (2007) argue for the existence of a threshold in seagrass density below which the seagrass engineering function is severely reduced, thus causing high turbidity and poor light penetration. Hence, as which the shift to the algae regime, this can also be reduced to a threshold regarding the light availability. Such collapse has been observed in the seagrass Zostera marina in areas around the north Atlantic in the 1930's where extensive die-off of seagrass due to an outbreak of the wasting disease caused a shift to a barren sediment regime (Duarte et al. 2006). In the Dutch Wadden Sea the meadows have not yet recovered, which is believed to be due to a synergistic effect between the disease outbreak and subsequent eutrophication (van der Heide et al. 2007).

The barren regime is reinforced by the reversed feedback loop that is created by the seagrass engineering function and otherwise maintains the seagrass regime. Without seagrass cover currents are not attenuated and sediments and nutrients are resuspended, which cause increased turbidity and light conditions below seagrass tolerance. It is therefore extremely difficult for seagrasses to recolonize once this regime is established (Cardoso et al. 2004; van der Heide et al. 2007). Furthermore, the increased current velocity renders a mobile and coarse bottom substrate that undermines seagrass root attachment, which can result in any new seedlings or remaining plants becoming uprooted (Cardoso et al. 2004). Surviving seedlings are also prone to fouling by epiphytic algae, as grazing pressure is low in young seagrass patches since they provide poor refuges from predators, and this further inhibits seagrass re-establishment (Valentine and Duffy 2006).

Uncertainties regarding climate change and regime shifts

The aspects of climate change that have been shown to affect regime shifts in seagrasses are sea level rise, increased CO2 and temperature rise (Short and Neckles 1999, Duarte 2002). But the connection to a regime shift remains speculative due to a lack of research (Short and Neckles 1999) and the fact that the outcome of climate change is co-dependent on other human activities in marine areas (Borum et al. 2004); most effects will vary spatially and depend on species in question. Temperature rise will affect photosynthesis in both algae and seagrasses and will depend on species thermal preference, but it has been shown that epiphytic algae growing on eelgrass will be favoured by higher sea temperatures (Short and Neckles 1999). Sea level rise will result in seagrasses losing their habitat and being forced to move in order to regain the light conditions needed and in addition the sea level rise will cause erosion that further will decrease light conditions for seagrasses (Short and Neckles 1999; Borum et al. 2004). An increase in CO2 levels might lead to an advantage for seagrass over algae since they require more CO2, but this is contested since evidence is weak (Borum et al. 2004). Most significantly, climate change will increase the risk for more extreme weather with more frequent and bigger storms which will cause sediment resuspension decreasing light conditions together with physical disturbance (Short and Neckles 1999; Duarte 2002).

Seagrass beds play a significant role in providing habitats and nursery grounds for marine organisms targeted for human consumption e.g. scallops, shrimps, crabs and juvenile fish (Duarte 2002; Terrados et al. 2004; Eklöf 2008; Barbier et al. 2011).  Thus, they are important habitats which enhance the welfare of people who directly are dependent on its resources (De la Torre-Castro et al. 2004).  If seagrass beds are lost due to a regime shift, its provisioning ecosystem services may diminish, which can be detrimental for the well-being of dependent people, especially affecting provision of food and livelihood. Fishermen, whose main source of income comes from seagrass-associated species, may be the user group most affected by the regime shift (De la Torre-Castro et al. 2004). Seagrass beds play an important regulating role by capturing carbon dioxide and transforming it into organic carbon (Duarte, 2002; Orth et al. 2006; Barbier et al. 2011). It has been suggested that the carbon stored in living seagrasses globally is on average 2.52±0.48 Mg C ha^-1 (Fourqurean et al. 2012). Seagrass bed decline could lead to a significant loss in the CO2 sequestration capacity and reduction in carbon storage, with potential negative effects at the global scale associated with climate change.

Seagrass beds attenuate waves and stabilize sediments and in doing so reduce coastal erosion and erosion of bottom substrates (Duarte 2002; Orth et al. 2006; Eklöf 2008; Barbier et al. 2011). They also can reduce the effects of storms and extreme weather events like hurricanes, providing a coastal protection service. It can be assumed that the lack of this service may affect the sense of security as material security of coastal populations and socio-economic activities in place such as fisheries, tourism, marine transport and aquaculture. In turn, fishing activity associated with seagrass beds benefit consumers by meeting food demand at local but also distant locations. In terms of access to enough nutritious food as a constituents of well-being (Reid et al. 2005), a regime shift may impact the adequacy of material for a good life of especially coastal communities. In addition, although in a limited way, some communities benefit from seagrass in that it is used as raw material and as food, as well as a fertilizer in some other regions (De la Torre-Castro et al. 2004; Barbier et al. 2011).Non-material services related to the aesthetic and cultural values of seagrass beds that benefits some traditional groups (Kenworthy et al. 2006; Barbier et al. 2011) could be lost if a regime shift takes place. Although tourism may threaten seagrass beds (Ochieng et al. as cited in Eklöf 2008) they could have an overall positive effect on the tourist industry (Duarte 2002) by providing an aesthetic setting with high water clarity and habitats for diverse species (Barbier et al. 2011). Therefore, as regime shift resulting in a reduction of these services could have a potential negative effect on the tourism industry.

One of the most crucial management actions is to limit nutrient input. Important measures for doing this are limiting the use of fertilizers in agriculture; protecting marsh areas, as they can act as a buffer against nutrient loading; treating wastewater (Duarte 2002) and regulating its disposal so that it is discharged in areas with efficient water exchange. These measures are also effective for reducing organic matter loading. Human-provoked physical disturbance should also be controlled. For example, management should limit dredging and marine constructions to areas outside seagrass beds when possible, and should limit dredging and sand reclamation to short periods that seagrasses can overcome (Borum et al. 2004). Management should also regulate fishing activity in order to avoid overfishing on top-predators and prevent cascading effects in the food-web, which otherwise can lead to algal dominance (Eklöf 2008). Efforts should also be made at an international scale to mitigate climate change (Borum et al. 2004). For an effective implementation of these measures, it would be necessary to increase public awareness about the ecological functions seagrass beds carry out and the services they provide to society (Duarte 2002; Orth et al. 2006; Eklöf 2008). 

Once seagrass ecosystems have shifted into a new state, recovery can be difficult or even irreversible in human time scale (Duarte 2002; van der Heide et al. 2007). Therefore, it is preferable to maintain or to build the resilience of these systems to prevent a regime shift, as restoring them once a shift has occurred can prove difficult if not impossible (Orth et al. 2006). In the event of a regime shift it is possible to return to a seagrass-dominated regime by resorting to seagrass transplantations. However these techniques have high costs and the success rates are low (Duarte 2002; van der Heide et al. 2007) therefore regime shifts are best prevented (Orth et al. 2006).