Much of the world's biodiversity is concentrated in nations with fragile governance systems exposed to repeated political crises1 that can threaten biodiversity2 and its associated benefits to people. Given recurrent political instability and ongoing biodiversity declines in many nations, there is an urgent need to identify which conservation interventions are most resilient during times of crisis and post-crisis recovery. Yet, there are no published studies of the relative performance of different kinds of conservation interventions during a political crisis. Here, we investigate how a political crisis affects the relative performance of community managed forests and state-managed protected areas.
Community forest management (CFM) and government-administered protected areas are among the most widespread conservation interventions around the globe. CFM has been promoted as an alternative to strict state-managed protected areas, to avert deforestation while also supporting the rights and interests of local people3. Identifying the conditions under which community-based or state management may be more effective at conserving biodiversity remains a key research question4,5. In more remote areas, local people may be better able to protect forests due to higher costs of centralized monitoring and enforcement6. Alternatively, state management may be more effective at conserving biodiversity if local institutions are weak or incentives for local conservation are insufficient4.
From a conservation perspective, there is evidence that local communities can conserve vulnerable ecosystems better than the state under certain biophysical, economic, cultural, or sociopolitical conditions3,7. For example, community forests were found to be more effective than state-managed protected areas in terms of reducing deforestation in Peru8. Community forests were effective at reducing deforestation relative to a counterfactual in India9 and Indonesia10 and at reducing forest disturbance in Tanzania11. A systematic review found that decentralized systems of forest management reduce deforestation, on average, but the effects are small12.
Evaluating the performance of interventions requires eliminating rival explanations for observed outcomes13. For example, many protected areas are established in remote areas or in areas that are unsuitable for agriculture, and therefore are unlikely to experience deforestation even in the absence of protection14. This makes it challenging to isolate the effects of different conservation interventions from other factors, such as remoteness. For example, multiple-use protected areas in Bolivia, Costa Rica, Indonesia, Thailand and Brazil were found to be just as effective, or more effective, than strictly protected areas at avoiding deforestation because they are more likely to be located in areas with higher deforestation pressures, such as closer to roads and cities15,16. The challenges associated with evaluating the relative performance of different interventions compromise our ability to test assumptions and design impactful conservation strategies.
A growing number of studies attempt to define what would have happened under a counterfactual scenario, in an effort to isolate causal effects of interventions8,15–17. A systematic review of 68 such studies found that estimates of the effectiveness of protected areas in terms of avoided deforestation were much smaller when counterfactual methods were used, compared to traditional methods18. A second review of 82 counterfactual-based studies found that protected areas were only moderately effective at reducing deforestation, on average, since they are typically placed in areas with lower pressures17. Other interventions, such as decentralized forest management and Indigenous protected lands, were more effective, but the number of studies using counterfactual methods were very small (three studies in each case.) In Peru, for example, indigenous territories and locally-managed conservation concessions were on average more effective than state-managed protected areas in terms of avoided deforestation and degradation, after controlling for confounding factors such as distance to roads and settlements8.
There have been very few counterfactual-based studies which investigate the effectiveness of conservation interventions in times of crisis. The few examples we identified focused on armed conflict. In Nepal, local institutions were able to organize and cooperate to reduce forest fragmentation even during periods of violent conflict19. In Colombia, large protected areas were more effective at reducing deforestation during periods of conflict between the government and guerilla fighters20. In Sierra Leone, armed conflict was linked to lower rates of deforestation, but the performance of conservation interventions was not specifically analyzed21.
There are also very few counterfactual-based studies which have assessed how sociopolitical context can influence conservation performance. Abman22 showed that deforestation rates were higher in less democratic nations that failed to control corruption or protect property rights. In Indonesia, direct elections boosted the ability of protected areas to prevent deforestation, but not forest fragmentation or fire23. Elections were found to increase deforestation in Brazil 24 and increase forest fires in Madagascar25. We know little, however, about the relative effectiveness of different conservation interventions during and after a political crisis.
Here, we investigate how a political crisis affected the relative performance of community-managed versus state-protected areas in the island nation of Madagascar. We define “performance” as the relative ability of conservation to reduce deforestation, after controlling for differences in location and other confounding variables.
Our evaluation focused on forests within Madagascar, a global biodiversity hotspot containing some of the most unique and threatened species on the planet26. Due to high levels of poverty and food insecurity, much of the island’s population depends on natural resources, including forests, for their livelihoods. An estimated eighty percent of Madagascar’s people live under the extreme poverty rate of USD $2.15/day and 40% of children under the age of five suffer from stunting27. The country lost 44% of its natural forest cover over the period 1953-201428 due to logging for timber and clearing for subsistence agriculture. A more recent analysis indicates that between 2000 and 2020 the country lost 4.85 million hectares, or 25% of its remaining tree cover29.
Madagascar’s government, often with support from international non-governmental organizations (NGOs) have attempted to slow deforestation through the creation of state-managed protected areas. As of 2020, the protected areas system included 110 sites encompassing 10.4% of Madagascar’s land area (6.1 million hectares)30. From 1990 to 2010, Madagascar’s protected areas were found to be effective at reducing deforestation, on average, but performance varied across time and space31. In northeastern Madagascar, for example, the establishment of new protected areas initially exacerbated ongoing deforestation, but later reduced rates of forest loss32. Protected areas in Madagascar are managed by different government agencies and NGOs. Here, we focus on 45 protected areas administered by Madagascar National Parks (MNP), a parastatal organization (Fig. 1). MNP sites are managed primarily for biodiversity conservation and restrict most human activity other than recreation.
In the 1990s, the Madagascar government instituted legislation that allowed for the creation of CFM contracts33. Contracts are established between a local forest management group (often supported by a non-governmental organization), the federal forest department, and in some cases, the local government. The terms of CFM contracts vary, but they typically prohibit new forest clearing and allow local use of renewable forest products for medicine, firewood, and food34. As of 2014, there were over 1000 CFM sites in Madagascar encompassing more than 3.1 million hectares, or 15% of the nation’s natural forests35. Previous research found that CFM had no detectable impact on deforestation, on average, between 2000 and 2010, but contracts that prohibited commercial use of forest products did reduce deforestation36. We focus on 362 CFM sites established prior to 2005 (Fig. 1).
Madagascar has experienced repeated political crises throughout its history. The most recent and prolonged crisis took place from 2009–2014, initiated by a global spike in rice prices, a large, surreptitious land deal between the government and a South Korean company, and frustration over corruption and oppressive governance37,38. Social unrest and political pressure led then-President Marc Ravalomanana to flee the country and an opponent, Andry Rajoelina, took power. The international community condemned the takeover as unconstitutional, and cut or eliminated foreign aid and investment39, causing a severe economic crisis. The crisis ended in January 2014, after democratic elections were held.
The combined political and economic crises impacted Madagascar’s forests and biodiversity. There was a spike in illegal logging of precious hardwoods such as rosewood40,41. Even within protected areas, illegal and extralegal logging took place as a result of limited capacity of park staff amd confusion caused by shifting regulations, in some cases with government permission or even cooperation41–43. Logging within national parks alarmed the international conservation community, which was concerned about potential extinctions of Madagascar’s lemurs and other unique species2.
At the same time, the crisis likely exacerbated pressures on community forests as local people sought to replace lost income from trade and tourism. Combined with already high rates of poverty and food insecurity, the crisis might have driven local people to clear forests to plant staple crops and meet their basic needs. Political unrest can also create conditions for increased deforestation. In Madagascar, there is evidence of excess forest fires set deliberately to protest outcomes of political elections25.
Given that Madagascar’s forests faced concomitant pressure from government dysfunction, political protest, and economic stress, we explore the effect of the 2009 political crisis on the relative performance of community forest management areas (CFM) and protected areas administered by Madagascar National Parks (MNP) during and after the crisis. “Performance” was defined as the ability to reduce deforestation, after controlling for differences in location and other potentially confounding factors. Our study builds upon prior work that evaluated the overall effectiveness of protected areas 1990-201031 and CFM 2000-201036, though without reference to political crisis. As such, we had no clear a priori predictions of which type of area would perform better amid the crisis.
To allow causal interpretation of our results, we used deforestation data derived from remote sensing28,44 and a counterfactual approach implemented through a combination of statistical matching and an event study design. We combined two methodological approaches to disentangle the effect of management type from other spatial and temporal factors that are known to influence deforestation. First, we used statistical matching to identify equivalent forest areas within CFM and MNP that are similar across a range of biophysical and geographic characteristics, such as remoteness and suitability for agriculture. Second, we conducted an event study analysis to control for relevant time-variant factors, such as rice prices and climate variables, as well as differences in deforestation trends in the pre-crisis period. We also explored the effects of spatial resolution on our results. Lastly, because the impacts of the crisis on CFM performance may vary due to differences in location or other characteristics, we explored the influence of potentially moderating factors such as distance to cities, distance to roads, and population density. In our study, the “event” is the onset of the crisis, and our specific research question is “What was the effect of the crisis and post-crisis period on relative performance of CFM and MNP, in terms of their ability to reduce deforestation?”