1 Adding Fuel to the Fire; the Effect of Political Unrest on Forest Burning in Sub-Saharan Africa 1.0 Introduction A third of annual anthropogenic CO2 emissions can be attributed to global biomass burning (Bird & Cali 1998). Forest fires in Indonesia in 1994 released between 0.81 – 2.7 Pg carbon; equivalent to 13 – 40% of annual global carbon emissions (Page et al., 2002). 90% of all fires in the Miombo Woodlands, Tanzania, can be attributed to anthropogenic factors including; agriculture, logging, charcoal burning, and arson (Edwin, 1998). Fire is reported to negatively impact 1% of the globes forest; however this number is underreported, especially in Africa (FAO, 2010). The natural factors affecting a fires regime include: the presence of herbivores, the ignition effect of a lightning strike, climate, and vegetation type, for example coniferous species are usually less damaged by fire events (Seedre et al., 2011). Some forests, i.e. boreal, burn in a cyclic manner, and many species rely on this burning to flourish. Less than 10% of all forest burning is cyclic, and the rest of global fire is classified as wildfire by the FAO (2010). Anthropogenic factors have a significant impact on a fires regime, agricultural practices such as slash and burn, the spread of the road network leading to deforestation and land use change around the border of the road (Nepstad et al., 2001), political unrest (Kull, 2004) and population density can all have an impact on the fire regime. The connection between forest fires and political unrest has been noted throughout the literature (see Thomas, 2012, Kull, 2002 and Kuhlken, 1999) and in a number of countries including; Malawi, Madagascar, Kenya and Ethiopia, but never before has it been quantified. Political unrest can often manifest itself in the form of protest or armed conflict, these can involve fire setting. This investigation evaluates and visualises the correlation between these two parameters by linking the MODIS burnt area product with World Bank political unrest indicators and the Armed Conflict Location and Events Dataset (ACLED) across sub-Saharan Africa. 1.1. Political Unrest
13
Embed
Adding Fuel to the Fire; the Effect of Political Unrest on ... · 1 Adding Fuel to the Fire; the Effect of Political Unrest on Forest Burning in Sub-Saharan Africa 1.0 Introduction
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Adding Fuel to the Fire; the Effect of Political Unrest on
Forest Burning in Sub-Saharan Africa
1.0 Introduction
A third of annual anthropogenic CO2 emissions can be attributed to global biomass burning
(Bird & Cali 1998). Forest fires in Indonesia in 1994 released between 0.81 – 2.7 Pg carbon;
equivalent to 13 – 40% of annual global carbon emissions (Page et al., 2002). 90% of all fires in the
Miombo Woodlands, Tanzania, can be attributed to anthropogenic factors including; agriculture,
logging, charcoal burning, and arson (Edwin, 1998). Fire is reported to negatively impact 1% of the
globes forest; however this number is underreported, especially in Africa (FAO, 2010).
The natural factors affecting a fires regime include: the presence of herbivores, the ignition
effect of a lightning strike, climate, and vegetation type, for example coniferous species are usually
less damaged by fire events (Seedre et al., 2011). Some forests, i.e. boreal, burn in a cyclic manner,
and many species rely on this burning to flourish. Less than 10% of all forest burning is cyclic, and
the rest of global fire is classified as wildfire by the FAO (2010). Anthropogenic factors have a
significant impact on a fires regime, agricultural practices such as slash and burn, the spread of the
road network leading to deforestation and land use change around the border of the road (Nepstad
et al., 2001), political unrest (Kull, 2004) and population density can all have an impact on the fire
regime.
The connection between forest fires and political unrest has been noted throughout the
literature (see Thomas, 2012, Kull, 2002 and Kuhlken, 1999) and in a number of countries
including; Malawi, Madagascar, Kenya and Ethiopia, but never before has it been quantified.
Political unrest can often manifest itself in the form of protest or armed conflict, these can involve
fire setting. This investigation evaluates and visualises the correlation between these two
parameters by linking the MODIS burnt area product with World Bank political unrest indicators
and the Armed Conflict Location and Events Dataset (ACLED) across sub-Saharan Africa.
1.1. Political Unrest
2
Political unrest will be constantly referred to in this study, and is being defined as “the
likelihood that the government will be destabilized or overthrown by unconstitutional or violent
means, including domestic violence and terrorism” (Kaufmann et al., 2010).
In sub-Saharan Africa, violent conflict, and other illegal activities have occurred in forested
areas where the government has failed to invest; these illegal activities have led to a decline in
investment in these areas, creating a positive feedback. The desire to control natural resources,
including timber has also been lead to conflict. It has been noted that during some conflicts, revenue
earned from the selling of forest resources has funded more warfare. Countries including Angola,
Burundi, Democratic Republic of Congo and Somalia, have all experienced violent conflict. The
conflicts in these, and many other countries in sub-Saharan Africa, have occurred in rural areas,
with poor road networks in order for the rebels to distance themselves from government
authorities (Kaimowitz, 2003).
1.2. The relationship between political unrest and forest fire
The relationship between forest fire and political unrest is an incredibly complex one and in
order to begin to understand the relationship it is important to explore the mental processes
behind deliberate fire setting. Scott (1985) described arson as a “weapon of the weak” used by
“relatively powerless groups” to protest against “those who seek to extract labour, food, taxes, rents
and interest from them”. Thomas (2012) describes “sociocultural fire setters”, who set fires as part
of a civil unrest movement to publicise their cause, the majority of these arsonists are noted to be
male.
Hoffmann et al., (2009), elaborates on this relationship, he notes that arson is the most
difficult type of anthropogenic fire to identify as its origin will often be disguised, instead of “arson”,
these fires will be classified as “accidental”, which makes specific analysis of arson events very
difficult. It is suggested that there is a strong link between forest fires and conflict over resources
and land use, this relationship has also been linked with underlying ethnic tensions (Kaimowitz,
2003).
Kocsis (2002), noted that people often commit arson attacks due to animosity; “instead of
an offender physically assaulting an individual with whom he or she has a grievance, the offender
may attack the victim (or their property) and use fire as their weapon”. This has been witnessed
with plantation workers in Malawi, who, when made redundant from their positions, set the
plantation alight (Mackennes, Pers. Comm., 2012). Arson has also been linked with the political
3
objectives of the individual, used as a form of violent protest including; terrorism, racial
discrimination and social/racial tensions. This link has been noted in a number of cases, notably in
Kenya in the 1920s. Protests against European colonists involved large forest fires and arson
attacks on the settlers’ crops. Another example of this was Algeria in 1956; it is believed that
nationalist rebels lit large fires in the Bainem State Forest (Kuhlken, 1999). Laris & Wardell (2006)
describe rural residents of West Africa lighting fires as a response to the governments’
implementation of policies restricting the use of slash and burn techniques.
Kull (2002) describes the problems in Madagascar. On the island fire setting is illegal due to
the damage caused to the islands natural resources; many people however rely on fire to support
their farming techniques. This clash of interests has seen a rise in the number of protests, and these
protests have often been in the form of fire lighting. In Madagascar, rural protest has been
identified as one of the greatest contributors to grassland fires. The Malagasy pop singer Rossy
released a song entitled “Resa Babakoto”, the lyrics of which encourage farmers to burn the hills to
attract the attention of the government to their struggle; Rossy disagreed with the billions of dollars
being spent on nature and conservation, while the Malagasy people were hungry and poor.
As sub-Saharan Africa has seen a lot of political and civil unrest over the past twenty years
(Makumbe, 1998) due to economic hardship and political regression (Bratton & van de Walle,
1992), it is ideally suited for an investigation of this kind. Much of the conflict seen in southern
Africa has been due to water scarcity; almost all water basins are shared between countries, this
creates conflict potential, both within the country, where people need access to a water basin, but
also between countries. It is thought these conflicts are exacerbated by underlying racial conflict
(Cochrane, 2009). This issue is mitigated via treaties, water management institutions and water
policy (MEA, 2005).
2.0 Methodology
This paper is composed of an investigatory analysis into the relationship between political
unrest and forest fires in Sub-Saharan Africa.
2.1. Burned area data
The Moderate Resolution Imaging Spectroradiometer (MODIS) fire product being used is the
burned area product, which displays the extent of burn scars over monthly periods, as opposed to
4
the active fire product which display the locations of fires occurring at the time of data collection
(Justice, 2002). The data being used in this analysis is at a 500 x 500m resolution.
2.2. Political unrest data
Two forms of political unrest data were used in this investigation; World Bank political unrest
indicators, and Armed Conflict Location and Event Dataset political unrest points. The World Bank
political unrest indicator is one of several indicators from the World Governance Indicators scheme
(WGI) and is on a country by country basis, with no specific spatial locations. The index being used
for this analysis is the political stability, no violence indicator. The data ranges from 1996 to 2010,
but does not include 2003 or 2011, so these years have been removed from the analysis, and as the
burnt area product is only available from 2002, it was chosen as the start point for the analysis.
These data are compiled from 30 underlying data sources detailing the perceptions of governance
of a large number of survey respondents and expert assessments, examples of some data sources
used to build the political unrest indicator include; EIU armed conflict, HUM, frequency of political
killings, disappearances and tortures, the IJT security risk rating, the IPD violent actions by
underground political organizations, the PRS internal conflict and ethnic tensions (Kaufmann et al.,
2010). The indicator itself officially ranges from -2.5 (weak) to 2.5 (strong) government
performance, however in many cases the political unrest indicator may reach as low as -3.9
(Kaufmann et al., 2010).
The second political unrest data set is the Armed Conflict Location and Event Dataset (ACLED).
These data contain specific dates and locations for different conflict events within Africa. Data types
include; violence against civilians, riots/protests, battles and rebel groups gaining territory. The
data are compiled from a range of sources, the main source being reports from war zones,
humanitarian agencies and other research sources. As all of the events in this dataset have XY
locations, they can be mapped (Raleigh, 2010), the points are not precisely spatially located, as they
are not derived from GPS data, but from reports, however, the general locations are likely to be
accurate as the data are designed to be used for mapping.
5
Figure 1; the methodology process is detailed above for Sudan in 2009;
a. MODIS burnt area data, b. MODIS burnt area data + ACLED political unrest points, c. political unrest points buffered by 50km, d. burnt area data extracted using these buffers
This method was applied to all countries and all years.
6
2.3. Forest fires in the vicinity of political unrest events
The null hypothesis for this investigation is that “there is no correlation between World Bank
political unrest indicators and burned area in Sub-Saharan Africa”, in order to disprove this, the
following method was used; each ACLED political result point was buffered by 50, 75, and 100km.
These distances were based upon;
a. the conceivable distance humans could disperse after a political unrest event
b. the distance between towns in Africa, which are generally far apart
c. the XY locations of the ALCED points are not precisely spatially located, so these buffers also
account for error in these locations.
It was decided that only fire within the buffers would be correlated with the World Bank
political unrest indicators. This method was chosen in an attempt to remove all fires from the data
which would not be caused by political unrest, as they occurred a significant distance away from a
reported event. The fires outside the buffers were deemed more likely to be naturally occurring, or
set as part of a farming technique. The fires within the buffers were more likely to be linked to the
political unrest event which had occurred, as they are closer.
The buffers were used to extract the burned area data surrounding the ACLED political unrest
points, see fig. 1, for each country. These rasters were used to calculate the total burning in the
country which occurred within the set distances (50, 75 and 100km) of an ACLED point by
comparing them with the total burning that occurred within that country that year. The results of
this analysis were correlated with the appropriate World Bank political unrest indicators. The
results were statistically tested using a Spearman’s rank correlation co-efficient.
3.0 Results
Fig. 2 illustrates the relationship between total burned area on a country by country basis,
and the World Bank political unrest indicator averaged over the study years. When the ACLED
buffers are not used, there is no relationship (rs = .052, p = .738). This may be due to the fact that
there are so many other factors influencing the total percentage of burning within the country. Fig.
3 illustrates the relationship between political unrest and the percentage of burnt area 50km from
an ACLED political unrest point averaged over the study period. The results in general show that
countries which have the highest amount of burning within the predefined buffers also have the
lowest political unrest indicators, they are politically unstable. The reverse of this has also been
7
Figure 2; the relationship between the total percentage of burnt area within each country and the World Bank political
unrest indicator for every country in sub-Saharan Africa averaged across 2002, 2004-2001. rs = -.052 and p = .738;
Spearman’s rank was not significant for this data set. The top 5 smallest countries are denoted by a triangle.
shown; countries which have the lowest amount of burning within the predefined buffers have
much higher political unrest indicators, they are more politically stable.
Figure 3; the relationship between the total percentage of burnt area within 50km of a political unrest event and the
World Bank political unrest indicator for every country in sub-Saharan Africa, averaged across 2002, 2004-2001. rs = -
.509 and p = .000; Spearman’s rank was significant to .001 for this data set. The top 5 smallest countries are denoted by a
triangle.
8
A spearman’s rank performed on all of the countries averaged over the eight study years to
test the relationship between the percentage burnt area 50km from an ACLED political unrest point
and the World Bank political unrest data was significant to the .001 level with rs = -.509 and p =
.000. The spearman’s ranks performed for each buffer (50, 75 and 100km) for each year being
studied were all significant, with p = .000 - .017. The result means that for all years and all buffers,
there is a significant relationship between the percentage of burning occurring within 50, 75 and
100km of an ACLED political rest point and the World Bank political unrest data.
A spearman’s rank was also performed on the burned area occurring outside of the political
unrest buffers. The result was not significant (rs = -.066, p = .672). This test was performed to
ensure that the correct buffer sizes were used, and that the fires occurring outside of the buffers
were caused by other factors, not political unrest.
3.1. Case Study; Nigeria and the Democratic Republic of Congo (DRC)
The two countries selected for further investigation are Nigeria and the DRC, as they are
contrasting. The changes in Nigeria’s political unrest indicator mirrored its changes in burned area
within the ACLED political unrest buffers. In contrast, despite its low political unrest indicators, no
more than 10% of the DRC burned within the predefined buffers during the study period. See tables
1& 2 for a full break down of their changes over time in terms of percentage burning at all buffers
compared with the World Bank political unrest indicator for each year.
9
The anomalous nature of the DRC may be due to the environmental and political structure
of the country. Most of the country is covered in dense rainforest (GLC 2000), which means fewer
fires will occur, but also that fires which do occur, will be less likely to be picked up by the MODIS
satellite, due to the canopy. The DRC is one of the least politically stable countries in sub-Saharan
Africa, this may have led to an underreporting of ACLED political unrest points due to poor
communication infrastructure.
4.0 Discussion
The result of this analysis is that with an increasing political unrest indicator, there is also
an increase in forest fires within 50, 75 and 100km of a political unrest event. This result matches
the opinions of; Thomas (2012), who links “sociocultural fire setters” with civil unrest, Hoffmann et
al., (2009) who links forest fires with conflict over resources, Kocsis, (2002), who links fire setting
with the feeling of animosity, and finally Kull (2002), who studied forest fires in Madagascar and
noted the relationship.
The countries with the most burning, i.e. Burundi, Liberia and Somalia, and countries which
burnt the least, e.g. Sao Tome and Principe and Gabon, have very different compositions of land use,
climate, population density and size of road network. The parameter these countries have in
common are their political unrest indicators, this emphasises the presence of the relationship.
Attempts have been made to understand why people set fires as a form of political unrest,
this phenomena has been described by Scott (1985), Thomas (2012) and Hoffmann et al., (2009).
10
The consensus of the literature is that, generally, fires are started in protest when groups feel
powerless to stop changes affecting their way of life, people will also start fires to publicise a cause,
or in order to attack an individual through the destruction of their property (Kocsis, 2002).
4.1. Implications for fire and forest risk modelling
The results of the analysis demonstrate that there is a statistically significant relationship
between political unrest and forest fires in sub-Saharan Africa. These results will not only have
significant impacts on fire and risk modelling, but also forest management.
Forest fire modelling is vital for predicting where fire is likely to occur, and how widespread
and possibly damaging a fire will be. These models need to be as accurate as possible, and all of the
important parameters affecting the likelihood of fire need to be accounted for, in order for the
results to be used for decision making (Perry, 1998). As previously stated, political unrest is not a
parameter in any known fire model. As a significant relationship has been discovered between
political unrest and forest fire in sub-Saharan Africa, this World Bank political unrest parameter
should be incorporated into any fire modelling in this area.
4.2. Implications for forest management
This analysis has shown that political instability has a significant effect on forest burning in
sub-Saharan Africa. This result will have an impact on forest management schemes, these schemes
should aim to be “top down”, as if corruption and conflict are not addressed, then schemes to
protect forests in unstable countries are unlikely to succeed. Forest protection schemes should
perhaps work in tandem with development schemes, as the two have been proved, through this
analysis, to be interlinked.
5.0 Conclusion
In conclusion, this analysis has proved that there is a significant relationship between forest
fires and political unrest in sub-Saharan Africa. This correlation confirms the relationship which
has previously been noted in the literature, but has never before been analysed. The implications of
these results are significant, and will have impacts on both fire and risk modelling. The result
cements the fact that approaches to forest management must be “top down”, or work in
combination with other development schemes which aim to combat corruption and political unrest,
without taking these factors into account, forest protection schemes such as REDD, are
fundamentally flawed.
11
References:
Bird, M.I., Cali, J.A., 1998. A million-year record of fire in sub-Saharan Africa, Nature, Vol. 394, pp.
767-769.
Bratton, M., van de Walle, N., 1992. Popular Protest and Political Reform in Africa. Comparative
Politics, Vol. 24, No. 4, pp. 419-442.
Cochrane, M.A., 2009. Tropical Fire Ecology: Climate Change, Land Use and Ecosystem Dynamics.
Praxis, Chichester.
Edwin, N., 2002. Fire in miombo woodlands: a case study of Bukombe District Shinyanga, Tanzania, in:
Moore, P., Ganz, D., Tan, L.C., Enters, T., Durst, P.B. (Eds.), Communities in flames: proceedings of an
international conference on community involvement in fire management. Food and Agriculture Organization
of the United Nations, Thailand, pp. 117-118.
Food and Agriculture Organization of the United Nations, 2010. www.fao.org [Accessed Online]