Search

Ecologists have long studied patterns, directions and tempos of change, but there is a pressing need to extend current understanding to empirical observations of abrupt changes as climate warming accelerates. Abrupt changes in ecological systems (ACES)-changes that are fast in time or fast relative to their drivers-are ubiquitous and increasing in frequency. Powerful theoretical frameworks exist, yet applications in real-world landscapes to detect, explain and anticipate ACES have lagged. We highlight five insights emerging from empirical studies of ACES across diverse ecosystems: (i) ecological systems show ACES in some dimensions but not others; (ii) climate extremes may be more important than mean climate in generating ACES; (iii) interactions among multiple drivers often produce ACES; (iv) contingencies, such as ecological memory, frequency and sequence of disturbances, and spatial context are important; and (v) tipping points are often (but not always) associated with ACES. We suggest research priorities to advance understanding of ACES in the face of climate change. Progress in understanding ACES requires strong integration of scientific approaches (theory, observations, experiments and process-based models) and high-quality empirical data drawn from a diverse array of ecosystems. This article is part of the theme issue 'Climate change and ecosystems: threats, opportunities and solutions'.

Regime shifts are difficult-to-reverse transitions that occur when an ecosystem reorganizes around a new set of self-reinforcing feedbacks. Regime shifts are predicted to occur when the intensity of some exogenous driver variable, such as temperature, annual harvest rate, or nutrient addition rate, gradually approaches and crosses a threshold value, initiating a transition to an alternative state. However, many driver variables now change rapidly as presses or pulses, not gradually, requiring new conceptual frameworks for understanding and predicting regime shifts. We argue that identifying and controlling regime shifts in response to presses and pulses will require a greater focus on the duration, not just the intensity, of changes in driver variables. In ecosystems with slower dynamics, transitions to an alternative state can take years to decades and as a result, a driver press with an intensity capable of resulting in a regime shift over long time spans may fail to cause a regime shift when applied for shorter durations. We illustrate these ideas using simulations of local-scale alternative stable state models and preliminary evidence from long-term grazing and eutrophication experiments. The simulations also suggest that small changes in the duration of driver presses or pulses can determine whether an ecosystem recovers to its original state. These insights may extend to larger scales. In spatially extended simulations that included patchiness, spatial heterogeneity, and spatial connectivity, all patches recovered to their original state after shorter presses. However, once press duration exceeded a threshold, growing proportions of the landscape shifted to an alternative state as press duration increased. We observed similar patchy transitions in a catchment-scale experiment that reinstated frequent fires approximately halfway through a regime shift from grassland to shrubland, initiated by fire suppression. In both the local- and larger-scale models, the threshold duration needed to elicit regime shifts decreased as press intensity increased or when factors counteracting regime shifts weakened. These multiple lines of evidence suggest that conceptualizing regime shifts as an interactive function of the intensity and duration of driver changes will increase understanding of the varying effects of driver presses, pulses, and cycles on ecosystem dynamics.

State transitions are changes in ecosystem structure and self-reinforcing feedbacks that are initiated when an exogenous driver variable crosses a threshold. Reversing state transitions is difficult and costly. While some state transitions are relatively rapid, many take years to decades. Outside of theoretical models, very little is known about slower state transitions and how they unfold in time and space. We quantified changes in spatial variance as a mesic grassland ecosystem shifts to a shrub-dominated state, using long-term experiments and simulations that maintain grasslands with annual fires or initiate a state transition to shrub dominance by decreasing fire frequency. In the experiments, the susceptibility to state transitions varied substantially in space. In the less frequent fire treatment, some plots became shrub-dominated around year 20 and grass extirpations began in year 25, but a third of the plots were still grass-dominated in year 37. Variable rates of state transition resulted in increasing spatial variance of grass cover over time, whereas shrub cover variance decreased. In the annually burned treatment, grasses remained dominant and the spatial variance of grass cover declined. In a separate experiment, less frequent fires were maintained for 23 years and then switched to annual fires. The switch to annual fires occurred shortly after grass variance started to increase and a majority of these plots quickly returned to a grass dominated state. In simulations, spatial variance remained low and average grass cover was high under frequent fires. If fire frequency decreased below a threshold, the ecosystem transitioned to shrubland, with a transient increase in the spatial variance of grass cover during the transition between states. Synthesis. Spatial variability in the rate and susceptibility to state transitions is indicative of a system with a patchy spatial structure, high spatial heterogeneity and low connectivity between patches. Increases in spatial variance can serve as an indication that some patches have begun a state transition and that management interventions are needed to avoid widespread transitions. This is one of the first empirical examples where altering management after an increase in spatial variance prevented state transitions.

Ensuring food security requires food production and distribution systems function throughout disruptions. Understanding the factors that contribute to the global food system's ability to respond and adapt to such disruptions (i.e. resilience) is critical for understanding the long-term sustainability of human populations. Variable impacts of production shocks on food supply between countries indicate a need for national-scale resilience indicators that can provide global comparisons. However, methods for tracking changes in resilience have had limited application to food systems. We developed an indicator-based analysis of food systems resilience for the years 1992-2011. Our approach is based on three dimensions of resilience: socio-economic access to food in terms of income of the poorest quintile relative to food prices, biophysical capacity to intensify or extensify food production, and the magnitude and diversity of current domestic food production. The socio-economic indicator has large variability, but with low values concentrated in Africa and Asia. The biophysical capacity indicator is highest in Africa and Eastern Europe, in part because of high potential for extensification of cropland and for yield gap closure in cultivated areas. However, the biophysical capacity indicator has declined globally in recent years. The production diversity indicator has increased slightly, with a relatively even geographic distribution. Few countries had exclusively high or low values for all indicators. Collectively, these results are the basis for global comparisons of resilience between nations, and provide necessary context for developing generalizations about the resilience in the global food system.

Spatially diverse trends in population growth, climate change, industrialization, urbanization and economic development are expected to change future food supply and demand. These changes may affect the suitability of land for food production, implying elevated risks especially for resourceconstrained, food-importing countries.Wepresent the evolution of biophysical redundancy for agricultural production at country level, from 1992 to 2012. Biophysical redundancy, defined as unused biotic and abiotic environmental resources, is represented by the potential food production of 'spare land', available water resources (i.e., not already used for human activities), as well as production increases through yield gap closure on cultivated areas and potential agricultural areas. In 2012, the biophysical redundancy of 75 (48) countries, mainly in North Africa, Western Europe, the Middle East and Asia, was insufficient to produce the caloric nutritional needs for at least 50% (25%) of their population during a year. Biophysical redundancy has decreased in the last two decades in 102 out of 155 countries, 11 of these went from high to limited redundancy, and nine of these from limited to very low redundancy. Although the variability of the drivers of change across different countries is high, improvements in yield and population growth have a clear impact on the decreases of redundancy towards the very low redundancy category.Wetook a more detailed look at countries classified as 'Low Income Economies (LIEs)' since they are particularly vulnerable to domestic or external food supply changes, due to their limited capacity to offset for food supply decreases with higher purchasing power on the international market. Currently, nine LIEs have limited or very low biophysical redundancy. Many of these showed a decrease in redundancy over the last two decades, which is not always linked with improvements in per capita food availability.

Disturbances are ubiquitous in ecological systems, and species have evolved a range of strategies to resist or rebound following disturbance. Understanding how the presence and complementarity of regeneration traits will affect community responses to disturbance is increasingly urgent as disturbance regimes shift beyond their historical ranges of variability. We define "disturbance niche" as a species' fitness across a range of disturbance sizes and frequencies that can reflect the fundamental or realized niche, that is, whether the species occurs alone or with other species. We developed a model of intermediate complexity (i.e., a Goldilocks model) to infer the disturbance niche. We parameterized the model for subalpine forests in Yellowstone National Park (USA) adapted to infrequent stand‐replacing fires and included the three major tree‐regeneration strategies: (1) obligate seeders that rely on ex situ seeding into burned areas (non‐serotinous lodgepole pine, Pinus contorta var. latifola), (2) obligate seeders that depend on in situ seedbanks (serotinous lodgepole pine, Pinus contorta var. latifola), and (3) species that can resprout from surviving roots following fire (quaking aspen, Populus tremuloides). Our results showed which regeneration strategies increase or decrease in prevalence as fire rotation declines. Non‐serotinous pines were extirpated when fire rotation was below 50 years in a monoculture and 100 years in a mixed forest; serotinous pines were extirpated when fire rotation was below 20 years; and aspen was extirpated when fire rotation fell below 6 years. The fundamental and realized disturbance niches pinpointed the key mechanisms limiting regeneration for each strategy, namely, increasing fire size for non‐serotinous pine (ex situ seeders), decreasing fire frequency for serotinous pine (in situ seeders), and interspecific competition for aspen (resprouters). In a mixed forest, the three regeneration strategies were complementary and each dominated at different combinations of fire size and frequency. Consequently, diversity of regeneration strategies enhanced forest resilience to declining fire rotations. Despite its simplicity, our Goldilocks model produced realistic dynamics and could be readily adapted to other disturbance‐prone ecosystems to explore the generality of these results. The disturbance niche is a key concept for anticipating community resilience to changing disturbance regimes. [ABSTRACT FROM AUTHOR]

Hysteresis is a fundamental characteristic of alternative stable state theory, yet evidence of hysteresis is rare. In mesic grasslands, fire frequency regulates transition from grass‐ to shrub‐dominated system states. It is uncertain, however, if increasing fire frequency can reverse shrub expansion, or if grass‐shrub dynamics exhibit hysteresis. We implemented annual burning in two infrequently burned grasslands and ceased burning in two grasslands burned annually. With annual fires, grassland composition converged on that of long‐term annually burned vegetation due to rapid recovery of grass cover, although shrubs persisted. When annual burning ceased, shrub cover increased, but community composition did not converge with a long‐term infrequently burned reference site because of stochastic and lagged dispersal by shrubs, reflecting hysteresis. Our results demonstrated that annual burning can slow, but not reverse, shrub encroachment. In addition, reversing fire frequencies resulted in hysteresis because vegetation trajectories from grassland to shrubland differed from those of shrubland to grassland. [ABSTRACT FROM AUTHOR]

Questions: Climate extremes are predicted to become more common in many ecosystems. Climate extremes can promote and interact with disturbances, but the combined effects of climate extremes and disturbances have not been quantified in many ecosystems. In this study, we ask whether the dual impact of a climate extreme and concomitant disturbance (wildfire) has a greater affect than a climate extreme alone. Location: Tallgrass prairie in the Konza Prairie Biological Station, northeastern Kansas, USA. Methods: We quantified the response of a tallgrass prairie plant community to a 2‐year climate extreme of low growing‐season precipitation and high temperatures. In the first year of the climate extreme, a subset of plots was burned by a growing‐season wildfire. This natural experiment allowed us to compare community responses to a climate extreme with and without wildfire. Results: In plots exposed to the climate extreme but not wildfire, community structure, diversity, and composition showed minor to insignificant changes, such as a 20% reduction in grass cover and a slight increase in species diversity. Plots exposed to both the climate extreme and wildfire underwent larger changes, including an 80% reduction in grass cover, 50% increase in forb cover, and increased plant diversity. Two years after the climate extreme, structural shifts in burned plots showed little sign of recovery, indicating a potentially lasting shift in plant community structure. Conclusions: Our results suggest that community responses to climate extremes need to account for climate‐related disturbances — in this case, high temperatures, drought and wildfire. This response diverged from our expectation that heat, drought, and an additional fire would favor grasses. Although growing‐season wildfires in tallgrass prairie have been rare in recent decades, they will likely become more common with climate change, potentially leading to changes in grassland structure. [ABSTRACT FROM AUTHOR]

Abrupt ecological changes are, by definition, those that occur over short periods of time relative to typical rates of change for a given ecosystem. The potential for such changes is growing due to anthropogenic pressures, which challenges the resilience of societies and ecosystems. Abrupt ecological changes are difficult to diagnose because they can arise from a variety of circumstances, including rapid changes in external drivers (e.g., climate, or resource extraction), nonlinear responses to gradual changes in drivers, and interactions among multiple drivers and disturbances. We synthesize strategies for identifying causes of abrupt ecological change and highlight instances where abrupt changes are likely. Diagnosing abrupt changes and inferring causation are increasingly important as society seek to adapt to rapid, multifaceted environmental changes. [ABSTRACT FROM AUTHOR]

This review synthesizes evidence that altered fire frequency drives discontinuous ecosystem transitions from mesic grasslands to shrublands or woodlands in the Central Great Plains, USA., Long-term fire manipulations reveal that grassland to shrubland transitions are triggered when fire-free intervals increase from 1-3 years to ≥ 3-8 years, and longer fire returns (˜10 years or more) result in transitions to woodlands. Grazing and soil properties alter these fire thresholds., Grassland to shrubland transitions are abrupt and exhibit nonlinear relationships between driver and state variables. Transitions to shrublands and woodlands exhibit hysteresis, where reintroducing frequent fires does not reverse transitions in management-relevant time-scales (decades)., Nonlinear transitions and hysteresis emerge because grasses generate positive feedbacks with fire that create strong demographic barriers for shrub and tree establishment. Fire-free intervals allow shrubs and trees to reach a size sufficient to survive fire, reproduce and disrupt the fire feedback loop through competition., Synthesis. Mesic grasslands, shrublands and woodlands constitute self-reinforcing states (alternative attractors) separated by critical fire frequency thresholds. Even without major shifts in climate, altered fire frequency can produce dramatic state changes, highlighting the importance of fire for predicting future ecosystem states. Local management should focus on prevention of unwanted transitions rather than post hoc restoration. [ABSTRACT FROM AUTHOR]

Intensified anthropogenic activities have drastically altered many ecosystems, motivating the use of restoration to regain key ecosystem functions and services and to stem biodiversity losses. Restoration is particularly difficult when human activities have pushed an area into a new state with self‐reinforcing feedbacks. This study investigated a long‐term restoration project in a dryland ecosystem of the Tengger Desert in northwestern China, initiated in 1956. We analyzed shrub and grass cover for 49 yr after the installation of restoration infrastructure that altered external conditions (i.e., using packed straw to reduce wind erosion) and system state (by planting shrubs). After 37–40 yr, the re‐vegetation project was successful in restoring the system to a state similar to native vegetation, with high grass cover (30–50%), low shrub cover (8–10%), and a thick biological soil crust (biocrust). However, the shift to high grass cover did not begin until year 37, before which shrub cover was high (15–20%) and grasses were subdominant (usually <10%). The shift from shrub to grass dominance was abrupt, registering significant nonlinear changes over time and relative to a key driver of vegetation dynamics, estimated biocrust thickness. Biocrust thickness increased gradually over time, which reduced rainfall infiltration into deep soil and thus increased soil moisture available for the shallow‐rooted grasses. The shift from the bare soil state to the steppe state exhibited a long time lag, suggesting that it can take decades to determine whether dryland restoration efforts succeed. The results indicate that persistence might be critical to forcing desired state transitions and that dryland restoration can proceed as a series of time lags, punctuated by abrupt changes in ecosystem state. [ABSTRACT FROM AUTHOR]