Emergent patterns in nature and society

Ebola: could disease outbreaks be a case of regime shifts?

Back in summer my colleague Victor Galaz (@vgalaz) asked on twitter what are the perceived global ecological risk further than climate change related disasters. To the questions, Professor Steve Carpenter replied that the risk of global pandemics, being one ebola, was one of his worries.   Today we face the IMG_2249biggest ebola outbreak in history. Some sources speculate that we will reach the 10.000 cases before the end of the year. While the media campaigns to better communicate the risk and reduce the panic, others try to make sense of the most recent data and update current models on the likelihood of spreading and size of infected population. One of the latest is the Mobs-lab lead by Alessandro Vespignani, a famous network scientist working amongst other things on disease spreading problems. Here is the link to the lab’s website where you can keep an eye on the latest datasets, a current issue on PLOS Current Outbreaks and their ‘live’ paper on ebola. The following video is another source of information about ebola statistics that takes into account the social side of it: cultural factors that affect the likelihood of transmission, poverty, communications issues and how our best hopes to contain the outbreak reside mainly on timely information and behavioural response.     As contribution to the broader discussion on how to deal with epidemics, not only ebola, I’d like to make available a draft for the Regime Shifts Database that I wrote back in 2009. Here I explored the possibility of conceptualise outbreaks as a type of regime shifts. The draft was never reviewed or published. Any comments or suggested readings to update it is more than welcome!

Outbreaks: from low to high frequency of events

  Summary of regime shift

Outbreak is a general term to refer an abrupt increase on a particular population size, often those that produce diseases in humans, animals or crops, and invasive species. Outbreaks could be both a regime shift in itself, or a driver of other regime shifts. In the first case, it happen when a threshold of susceptible individuals is passed. In the second case, the frequency of outbreaks can lead to chaotic-like dynamics: strange attractors, chaotic cycles and boundary coalitions. The change on the frequency of outbreaks could lead to structural change in ecosystems and society. Implications for ecosystem services includes loss of agricultural production, threats to human health, and landscape configuration.

Description of the alternate regimes and reinforcing feedbacks

Outbreaks are a regime shift by itself and a driver of other regime shifts (Scheffer, 2009). The first case is some times related to a shock event. There are events that happen once, for example in the case of diseases outbreaks a lots of victims die (human, animals or crops) and then the disease disappears. It could be that the host develop immunity or that the disease is so isolated that there is not chance of new infections that trigger a new outbreak. Other example is the one of invasive species outbreaks. In the absence of natural controls, these species hardly leave their new habitat establishing two regimes: with and without invasive species. However, this is a rare case because at some point the outbreaking organism (insect, parasite, virus, etc) ran out hosts, or the invasive species ran out resources. Then, a sharp decrease is followed by a dormant phase where the population is keep in low densities (Blarer & Doebeli, 1999). Thus, outbreak dynamics tend to be locked in strange cycles between two stable states: low abundance (healthy systems) versus extremely hight abundance (sick systems). Surprisingly, such dynamics can have periodic cycles and present synchronization (Stone et al., 2007). The frequency of flu increase in spring, measles on children increase at the beginning of the scholar year (Stone et al., 2007) whereas spruce budworm has outbreaks cycles every 40 years (Ludwig et al. 1978). Outbreaks might be a driver of other regime shifts. Knew examples include the shit from coral to algae dominance trigger by a disease outbreak on urchin barrens (Mumby et al., 2007), and the shift from primary to secondary arrangements in boreal forest (Scheffer 2009). Note that a second interpretation of “regime shift” might arise when the explanatory variable is the frequency of the strange cycle, rather than the healthy versus sick states described above. In such cases, small change in a parameter or a structural variable of the system in question can lead to an acceleration or slow down of the outbreak frequency. This effect might transform completely the structure, hence identity of the system. For example, insect outbreak cycles seems to be synchronized in northern American forests. The fragmentation of primary forest due to logging activities, settlements, or road construction might break down such synchrony. Consequently, different patches present outbreaks at different times. Depending on the landscape structure, in the new regime outbreaks could jump from patch to patch, making them more frequent in time and heterogeneous in space, exacerbating their economic consequences. Under such frequency regime, the forest might be trapped on a secondary state (trees are kept in youth ages, they never reach maturity) (Brassard & Chen, 2006). Another example is the prevalence of infectious diseases in poor countries. The poverty trap is explained by the following mechanism: under lack of enough income, people cannot afford protection against infectious diseases via nutrition or sanitary conditions; therefore they spend greater time infected and consequently produce less income (Bonds et al., 2009). In the later case, the structural variable (where the threshold is) would be some indicator of welfare in the country like income per capita. The mechanism underlying outbreaks are rather simple. An outbreak happen when a threshold of susceptible individuals is passed (Scheffer 2009). However, feedback mechanism become a little bit more elaborated when regarding specific cases. Environmental conditions and timing play a fundamental role when determining whether an outbreak is about to happen (Blarer & Doebeli, 1999; Stone et al., 2007). Although it is mathematically complicated to predict when an outbreak is going to happen, the timing plus the configuration of the population right after the last outbreak event (relative proportion of infected, recovered, susceptible and exposed individuals) could determine when the next event is probable and its magnitude (Stone et al., 2007). In addition, Blarer and Doebeli (1999) found that temporal coherence and the magnitude of population’s outbreaks is maximized at intermediate levels of environmental noise. Thus, at least in some cases, evidence suggest that the frequency of outbreaks is determined by its history. In the same line of reasoning, Peterson (2002) examine the effect of ecological memory in the emergence of landscape patterns due to contagious disturbances. Although he looks specifically at fire dynamics, outbreaks are contagious disturbance processes as well. His findings support the idea that some system with high memory could exhibit persistent patterns determined by the frequency of the disturbance and the structure of the system. Another possible mechanism is basin boundary collisions (Vandermeer & Yodzis, 1999). Briefly, it refers to cases where the regime shift is provoked by a subtle change in the boundary of the basin of attraction, getting close the attractor and the separatrix with adjacent basins. In other words, structural change in the system change the point where it might jump from one regime to another, rather than changing the dynamics. Such is the case of the use of pesticides that can induce mortality of natural enemies. Outbreaks would tend to be more frequent and present chaotic transients – a sort of flickering effect when the system goes to the alternative state and comes back repeatedly before actually shifting to second regime – (Vandermeer & Yodzis, 1999). Another example is vaccination, exemplified by Janssen (1998) in a model where malaria can evolve drug resistance. Following a period of low incidence, malaria picks up due to the increase in susceptible individuals against resistant varieties. Therefore there is two threshold to bear in mind. First of all, the amount of susceptible individuals in the target population. Second, structural variables that may affect the frequency of the outbreak. The second class would depend on the system under consideration.

Drivers that precipitate the regime shift

Outbreak is a general term to refer an abrupt increase on a particular population size, often those that produce diseases in humans, animals or crops, and invasive species. Consequently, the drivers of the regime shift covers a broad set of categories. When looking at the regime shift in its most simple way, healthy versus sick states, all that is needed is to cross a threshold on the susceptible individuals (Scheffer 2009). Therefore, the income of new individuals would make the population prone to new outbreaks. This includes population growth and migration dynamics as drivers. The loss of biodiversity can drive outbreaks under the same threshold mechanism. Thus, when predators or competitors densities are reduced both due to extinction or overexploitation, other species become abundant, hence prone to outbreaks. Such is the mentioned case of sea urchins in the Caribbean (Bellwood et al., 2004) In addition, hight connectivity could facilitate the spread of outbreaks (Norberg and Cumming, 2008). Under this category can be grouped the increase of trade facilities, expansion of transport systems and the concentration or establishment of urban environments (Foley et al., 2005). On the other hand, when regarding outbreaks regimes as changes in their frequency that are able to transform the system where they are produced (e.g. poverty traps, secondary forest); drivers become more subtle. Connectivity again play a fundamental role. Such is the case of landscape patterns that facilitate or not the spread of contagious disturbances (Peterson, 2002). Additionally, environmental oscillations like ENSO may trigger outbreak dynamics through stochastic resonance mechanism, that is a nonlinear interaction between environment and organism response (Blarer & Doebeli, 1999). Climate change is thought to increase the risk of pest outbreaks, including malaria, cholera, dengue, hantavirus pulmonary syndrome, influenza, and diarrhea among others(Ford et al., 2009; Khasniss & Nettleman, 2005; Janssen, 1998). Thus, not only connectivity in space but in time can modulate the frequency of outbreak dynamics. The introduction of pesticides or intensive vaccination programs might generate basin boundary collisions (Vandermeer & Yodzis, 1999; Janssen, 1998). A system with low frequency outbreak is then attracted to a high frequency state by the subsequent increase of susceptible individuals. In social systems, the lack of infrastructure and education may trap societies in regimes where outbreaks are more frequent. As a result, this societies become poorer and less capital for infrastructure and education is produced. Then poverty is a driver of high frequency outbreaks in human systems (Bonds et al., 2009; Khasniss & Nettleman, 2005).

Impacts on ecosystems and human well-being

Outbreaks are part of ecosystem dynamics, some species have evolved to present periods of extremely high densities and periods of dormancy (Blarer & Doebeli, 1999). In this sense, outbreaks represent intermediate pulse-like disturbances ecosystems are used to deal with (Scheffer et al., 2008). Most of the cases outbreaks works as a control for species with high densities. However, their frequency and intensity might modulate the ecosystem where they are present, favoring a set of species over other. Thus, outbreaks might be linked with other regime shifts where change in species dominance is a determining factor. Since outbreaks affect all organism, ecosystem services related to food production both crops, livestock, fisheries and wild animals are affected. Timber production, woodfuel and other products forest-related are included as well. Strange and Scott (2005) clearly capture the big picture for crops: 800 million people suffer from hungry, 1.3 billion live in poverty (less than US$ 1 per day) and at least 10% of global food production is lost due to plant diseases. The Great Bengal Famine back in 1943 caused 2 million people die due to high dependence on rice crops which suffer an outbreak collapse; whereas the southern corn leaf blight epidemic leave severe economical losses in the 1970’s US agricultural economy (Strange and Scott, 2005). Oerke (2005) estimated relative global losses due to pests for major crops as follows: from 26 to 29% for soybean, wheat and cotton, 31% for maize, 37% for rice and 40% potatoes. Human related diseases are not the exception. They follow the same dynamics and causal pathways as other outbreaks cases. Moreover, climate change is predicted to increase outbreaks intensity and frequency (MEA, 2005), including human disease outbreaks like diarrhea, malaria, dengue, influenza, cholera, and hantavirus (Ford et al., 2009).

Management options for preventing or reversing regime shift

Managerial options depend in case specific features of the regime shift including the type of organism, dispersion, vectors, and interactions with noise sources like climate variability. Given the drivers mentioned above, three generic managerial options arise. First, population growth and migration control. The likelihood of an outbreak increase in high density arrangements like large scales and homogeneous crops, intensive livestock production or big cities. Therefore, arrangements with low density might reduce this likelihood. Since connectivity plays a fundamental role when it comes to outbreaks, it is convenient to manage and monitoring all possible transportation links. In the human case, it comes to traveling facilities and migration movements. When it comes to food production, it has to be with landscape connectivity, vector migration dynamics and trade of resources. For example, biological corridors has been thought to be good barriers against dispersion of plants diseases or its vectors. In that sense, the maintenance of biodiversity both in the landscape and in food webs reduce the propagation of outbreaks dynamics. Second, policies for using vaccination or pesticides should take into consideration the evolutive dynamics, or the possibility of developing resistance, of both the host and the problematic organism (Janssen, 1998). When such dynamics are not taken into consideration, extensive use of vaccination and pesticides increase the amount of susceptible individuals to new resistant organisms. Then, the threshold of susceptible individuals is easily crossed and a bigger outbreak might happen. Third, monitoring programs would help to predict and react to outbreaks. Ford et al. (2009) suggest that the use of satellite images might help to predict forthcoming outbreaks by monitoring real time changes on landscape dynamics. Most of human related disease keep strong correlation with climate variables like temperature and humidity, landscape connectivity, water fluxes and its quality or environmental factors that affect their vectors behavior. In the same line of reasoning, information technology might help us to create early warning systems that allow us to react faster than the outbreak by preparing national health systems (Galaz, 2009; Ford et al., 2009). Finally, strategies that include reduction of poverty and empowerment might help to break the poverty trap. For example, education and sanitation technology have help to reduce significantly water-borne diseases in poor areas (Khasniss & Nettleman, 2005). However, other vulnerability factors like malnourishment, access to clean drinking water and local management of vector threats need to be addressed (Khasniss & Nettleman, 2005).


Blarer A,and Doebeli M. 1999. Resonance effects and outbreaks in ecological time series. Ecol Lett. 2 (3) pp. 167-177

Bellwood D, Hughes T, Folke C, and Nyström M. 2004. Confronting the coral reef crisis. Nature. 429 (6994) pp. 827-833

Bonds M, Keenan D, Rohani P, and Sachs J. 2009. Poverty trap formed by the ecology of infectious diseases. Proceedings of the Royal Society B: Biological Sciences  pp. 1-9

Foley J, DeFries R, Asner G, Barford C, Bonan G, Carpenter S, Chapin F, Coe M, Daily G, and Gibbs H. 2005. Global consequences of land use. Science. 309 (5734) pp. 570-574

Ford TE, Colwell RR, Rose JB, Morse SS, Rogers DJ, and Yates TL. 2009. Using Satellite Images of Environmental Changes to Predict Infectious Disease Outbreaks. Emerg Infect Dis. 15 (9) pp. 1341-1346

Galaz V. 2009. PANDEMIC 2.0 Can information technology save the planet?. Environment 51 (6) pp. 20-28

Janssen M. 1998. Use of complex adaptive systems for modeling global change. Ecosystems 1 (5) pp. 457-463

Khasnis A, and Nettleman M. 2005. Global warming and infectious disease. Arch. Med. Res. 36 (6) pp. 689-696

Ludwig D, Jones D and Holling C. 1978. Qualitative analysis of insect outbreaks systems – spruce budworm and forest. J. Anim. Ecol. 47 (1) pp. 315-332

Mumby P, Hastings A, and Edwards H. 2007. Thresholds and the resilience of Caribbean coral reefs. Nature  450 (7166) pp. 98-101

Oerke E. 2006. Crop losses to pests. J. Agric. Sci. 144 pp. 31-43

Peterson G. 2002. Contagious disturbance, ecological memory, and the emergence of landscape pattern. Ecosystems 5 (4) pp. 329-338

Scheffer M, Nes E, Holmgren M, and Hughes T. 2008. Pulse-Driven Loss of Top-Down Control: The Critical-Rate Hypothesis. Ecosystems 11 (2) pp. 226-237

Stone L, Olinky R, and Huppert A. 2007. Seasonal dynamics of recurrent epidemics. Nature 446 (7135) pp. 533-536

Strange R and Scott P. 2005. Plant disease: A threat to global food security. Annu. Rev. Phytopathol. 43 pp. 83-116

Vandermeer J and Yodzis P. 1999. Basin boundary collision as a model of discontinuous change in ecosystems. Ecology 80 (6) pp. 1817-1827


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s