Occurrence and Management of Wildfires in Northern Hhohho, Swaziland

A study was conducted to determine the status of wildfire and fire management practices in Swaziland. The northern Hhohho region was used as a case study. Information for occurrence of fires was sourced from MODIS fire alert system over a period of 12 months (August 2011 to July 2012). A proforma was used to collect data on land tenure, land use and cover and fire management and suppression measures from land users adjacent to 130 fire sites that were randomly selected from the case study area. Climatic data (rainfall and wind) were obtained from local weather stations. A total of 1,779 fires were recorded over the 12 months period in the whole country and 50% of the fires occurred during the months of August and September. The majority of fire spots (58%) were found in private land as opposed to 42% found in communal land. A significant difference for fire intensities (confidence levels) was observed for forest plantations and cultivation as well as for forest plantations and grazing (ρ<0.05). Honey hunters were the main causes of wildfire, with 29% of the fires attributed to them. This was more prevalent in the forest plantations though some forest plantations provided forest scouts to assist honey hunters to extract the honey, in an effort to reduce the incidents of wildfires. About 56% of wildfires were not suppressed and they caused the maximum possible destruction. The forest plantations had trained fire brigades and yet the rural communities did not have such and they lacked incentives to manage fires, especial when the fires were in adjacent private farms.


INTRODUCTION
Fires are mainly used to prevent abandoned The Kingdom of Swaziland is a monarchy and land-crops.They are also used in pest management to locked country located in the south eastern part of Africa reduce insects and rodents and to kill invader weeds covering an area of 17,363 km .It is bounded by the as well as in clearing vegetation to facilitate growth 2 Republic of South Africa on the north, west and south of grasses and tree sprouts to benefit animals.and by the Republic of Mozambique to the east.The Most communities use fire as a management tool at population of Swaziland was estimated at 1,067, 773 in the end of the dry season to promote new grass 2007, with one ethnic group called Swazi [1].In Swaziland, growth and for parasite control.Fires controlled or uncontrolled wildfires start in many ways; accidentally uncontrolled have profound impacts on the physical (e.g.misuse of appliances, dropping a cigarette or match environment including; land cover, land use, biodiversity, on a sofa or mattress, etc.), deliberate ignition or arson, forest ecosystem and climate change.The environmental equipment failure, including electrical malfunctions impact of fires includes loss of vegetation and deand overheating, honey collection as well as ritual forestation and loss of biodiversity.Wildfires have an performances, where candles are used and left burning.
impact on hydrological cycle where trees which act as Some fires start from neighbouring countries (South carbon sink are reduced and the smoke from these Africa and Mozambique) and spread over the border into wildfires contaminates the air.The socio-economic Swaziland.For example, in July 2007, fire that originated impacts include reduction of grazing lands, disruption of from South Africa and fanned by high winds destroyed critical infrastructure, destruction of power disruptions, 23,343 ha (76%) of Peak Timbers plantations and one deaths, homelessness and job losses in plantation timber mill was destroyed [2]. industries.
fields from growing over and clear areas for planting The factors that have influence on occurrence and owners and communities require an early warning system behaviour of wildfires are fuel, weather and topography to identify critical time periods of extreme fire danger [3].Usually, the greatest volume of fuel readily available periods in advance [13].The Swaziland Meteorological for burning the more intense the fire will be [4].When the Services, through e-mail, provides fire danger ratings as moisture content is high, fires are difficult to ignite and a tool for early warning of possible fire outbreaks in the burn poorly if at all [5].With little moisture in the fuel, country.The system gives a four day warning based on fires start easily and wind and other driving forces may prevailing weather conditions [14].The effective cause rapid and intense fire spread [6].Size and shape involvement of the local people has been recognised as a affects the surface area to volume ratio of fuels [7] hence prerequisite for the successful implementation of fire small fuels and flat fuels have a greater surface area to management programmes, especially at the interfaces volume ratio than larger fuels [8].Wind makes fire burn between wild lands, managed systems and residential faster by increasing the supply of oxygen and by driving areas [15].radiant heat into adjacent fuel [9].
The objective of the study was to determine the Predicting the potential behaviour and effects of status of wildfires and fire management systems in the wildfire is an essential task in fire management.
northern Hhohho region of Swaziland.Mathematical surface fire behaviour and fire effects models and prediction systems are driven in part by fuel Methodology bed inputs such as load, bulk density, fuel particle size, Description of Case Study Area: The study area is heat content and moisture of extinction [10].The MODIS bordered by Lomati River in the north and the Komati fire products build and improve on the experience of fire River in the south in the northern part of Swaziland assessment primarily using NOAA-AVHRR and GEOS (Fig. 1).It covers an area of 3,888 km .The study area satellite system [11].The fire products include an consists of communal land and private land covering an identification of the occurrence of thermal anomalies and area of 1,998 km and 1,883 km , respectively.The estimate the total emitted power from the fire or burned dominant land uses are communal grazing, ranching and area [12].To mitigate these fire-related problems, forest subsistence cropping and forest plantations.A small and land management agencies, as well as property portion (17 km ) is under wildlife conservation.Data for Fires: The author s subscribed to the NASA analysis using descriptive analysis (frequencies) and FIRMS (Fire for Resources Management System) e-mail comparison of means for the different parameters.The alert system that gives global location of fires on daily level of confidence of fire as reported on the satellite alert basis.FIRMS is based on MODIS (Moderate Resolution was taken as dependent variable and the factors which Imaging Spectroradiometer) system on board Terra and were considered to control the occurrence of fire were Aqua satellite systems [16].Data showing the location of moisture content, wind direction, wind speed, land tenure, fires within the country for 12 months period, from 1 land use and type of vegetation.One way ANOVA was st August 2011 to 31 July 2012 was used for the research.used to determine the significance in level of confidence st A total of 1,779 fires were recorded in the country during on reporting fires for wind speed, land use and type of this period.Information that was sourced from the e-mail vegetation cover.The LSD was used to separate alert included latitude and longitude of fire points, date significant mean differences.The Independent Samples Tand time of fire, satellite from which observations were Test was used to determine if there was a significant made and confidence level of fire detection.
difference in confidence level for different land tenure Selection of Sampling Sites: A total of 198 fires were found to have occurred within the case study area over a

RESULTS AND DISCUSSIONS period of 12 months. A sample calculator [17] was used to determine the ideal sample size at 95% confidence level
Distribution of Fires at National Level: The 1,779 fires and 5% confidence interval.The fires within the case spots that were recorded over the one year study period study area were allocated identification numbers and 130 are shown in Figure 2.They were denser on the western fire sites were randomly selected for field verification and part of the country and sparse on the southern part.Since collection of additional information from land users.
fires depend highly on climatic, topographic and fuel Data Collection: Information on climatic and weather in the Highveld because of higher altitude and steep conditions (rainfall, wind speed, wind direction) for the areas.The forest plantations were found mostly in the period when fires were reported was sourced from the eastern part of the country.The majority of the fires Swaziland Meteorological Services, Mhlume Sugar Estate occurred during the months of August and September and Peak Timbers Plantation.The coordinates for the (dry season) and confirmed other findings that wild fires sampled fire points acquired from NASA FIRMS were are the most widespread ecological disturbances in winter loaded into a Global Positioning System (GPS) and the and before the onset of the first rains when the forests GPS was used to navigate to the locations for data would be dry and highly inflammable in southern Africa collection.[21].During these months the vegetation would be dry Field Verification and Interviews: A proforma was winter season.Subsistence farmers also use fires to clear designed to collect data from the field for the selected land in preparation for cultivation and in anticipation of fires.The information collected from the field included the start of the rainy season.land tenure, land use and land cover.Respondents from Occurrence of fires in relationship to last rains and farms or homesteads within the vicinity of each fire site prevailing winds: The influence of rainfall on wildfire were asked to provide information on likely causes of the occurrence is due to its influence on fuel moisture and fire, effects of the fire, how fire was suppressed and load [22].About 43% of the fires occurred within a month constraints in management and suppression of wildfires.
after rainfall of less than 10 mm was received.As expected

Data Analysis and Presentation:
The data obtained from (wet periods).The study area recorded the highest the satellite e-mail alert was converted to spreadsheet number of fires between 0.0-9.9km/hr,having 99 (76%) of format and imported to ArcView GIS [18] for plotting maps the fires.The 20.0-29.9km/hr had the least number of fires showing distribution of fire spots both at national level recorded with 4 (3%) counts in the study area.There were and at the level of the case study area.The same data was 7 (5%) fire recorded when the wind speed was between further exported to SPSS software [19], were information 30.0 and 39.9 km/hr.Under atmospheric conditions on weather conditions and that collected from field suitable for controlled burning, wind plays a significant surveys were added.SPSS software was used for data but non-prevalent role in affecting fire intensity as studies systems.
factors, [20] the study confirmed that most fires occurred due to frequent winds that were experienced during the very few fires were recorded when heavy rains had fallen The study area was dominated by forest plantations and shrubs which reduces the wind speed.The fact that there were few fires during periods of high wind speeds may also be attributed to the fact that the wind speed of between 0 and 9.9 km/hr occurs more often in the study area than high wind speeds.The fires were recorded on all wind directions during the study period.Since Swaziland has rugged topography, the results obtained are in agreement with earlier studies that wind flows along the direction of hills and mountains [24].The dominant wind directions for occurrence of fire were the east, north-east and west.The least fires were reported when the wind direction was south-west and north-west.

Effect of Land Use and Land Tenure on Occurrence of Fire:
The majority of fire spots (58%) were found in private land as opposed to 42% found in communal land confirming that the high influence of land cover to wildfire activity had a significant effect on and close association with land use [25].About 46% of the fires occurred in grazing land (both communal grazing and private ranching).The study area recorded 28% of the fires in forest plantations (Table 1).Wildlife conservation covers about 0.5% of the total study area and yet about 4% of the reported fires were in wildlife conservation area.
There was no significant difference between fire intensity (confidence level) for private land and communal land ( > 0.05), even though the average for private land was higher at 69.88 (Table 2).The land tenure system in the study area had similar topography, climate and vegetation pattern.
There was a significant difference between fire intensities (confidence levels) for the different land uses ( < 0.05) (Table 3).The significance was observed between forest plantations and cultivation as well as between forest plantations and grazing (Table 4).The fire intensity was higher for forest plantations compared to cultivated land and grazing land.The cultivated land and grazing land had very little biomass during the period when wildfires were dominant (August to September).The overall effect of fire intensity was that grasses are generally not sensitive to increasing fire intensities than forests and bushes [26].
Reported Causes of Fire: Out of the 130 fire spots in the study site, 31 of them (24%) were intentional fire that was under control and 76% of the fire could be classified as wildfire.The controlled fires were in sugarcane fields (burning sugarcane before harvesting), forest plantations (fire breaks) and also burning of fields after clearing bush (slash burn).Honey hunters were the main source of wildfire, with 29% of the fires attributed to them.The honey hunters started fire with the intention of smoking bees and the fire got out of control.This was more prevalent in the forest plantations and some forest plantations had come up with a strategy where they assisted honey hunters to extract the honey.The located bee hives in their plantations were harvested by trained forests scouts in an effort to reduce the incidents of wildfires.Cigarette smokers also contributed in causing wildfire, especially on forest plantations, as they threw away burning cigarettes along the roads or foot paths.To reduce the destruction caused by burning cigarettes, forest plantations often make buffer zones by clearing vegetation within about 50 metres along the road or busy ------------------------------------ footpaths.Arson was reported to have contributed about 8% of the fires (Table 5).
Fire Management and Suppression: Fifty one wildfires caused the maximum possible destruction because they were not suppressed.This was a clear sign of national problems in suppression and management of wildfires.
The forest plantations had their own trained fire brigades that managed and suppressed fires.The brigades from forest plantations suppressed 23 wildfires.They were complemented by the Swaziland National Fire and Emergency in suppressing eight fires.The terrain in the study area, especially in areas used for forest plantations were very steep and in most cases not accessible by vehicles.Aircrafts and helicopters were sometimes used to suppress the fires.The aircrafts were hired from fires management companies in neighbouring South Africa and aircrafts from the air wing of the Swaziland Defence Force were sometimes involved in controlling the forest fires.
In communal land, the communities were responsible for management and suppression of wildfires and they managed to suppress 22 out of the 99 wildfires in the study area.

Fig. 1 :
Fig. 1: Map of Swaziland showing location of study area and ecological zones

Table 2 :
T-Test results for land tenure and fire intensity (confidence level)

Table 4 :
Multiple comparisons for means for different land uses Table 5: Reported causes of fire (N=130) *+ Mean difference is significant at the 0.05 level for respective land uses

Fire Management and Suppression:
The (such as the Forest Prevention Act of 1910 and Grass Fire Rural communities did not have trained fire brigades, Act of 1955) and they needed to be reviewed.unlikeintheforestplantations.They lacked training and knowhow in fire management and suppression.A lesson Fire as a tool in the management of cases in the study.The laws and legislation that intended a savannah/dry forest reserve in Madagascar. to assist in management and control of wildfires were Applied Vegetation Science, 2: 117-124.oftennotenforcedbecause of lack of capacity to do so 5. Stott, P., J.G. Goldammer and W.L.Werner, 1990.and also because of difficulties involved.Some of the The role of fire in the tropical lowland deciduous laws were out-dated (such as the Forest Prevention Act forests of Asia.In: J.G. Goldammer (ed.), Fire in the of 1910 and Grass Fire Act of 1955)[29, 30]and they Tropical Biota: Ecosystem Processes and Global needed to be updated.The existing interventions lacked Challenges (21-44).Berlin, Heidelberg.effective education to change public attitude, effective 6.IFFN., 2002.International Forest Fire News, laws and enforcement, adequate technical strategies and Fire Situation in India.Int.Forest Fire News, appropriate community capacity to deal with fires.Out of 130 fire spots that were used in the Land Cover Mapping, Fire Regeneration and Scaling case study, 58% occurred in private land as opposed to Studies in the Canadian Boreal Forest with 1-km 42% that occurred in communal land.The dominant wind AVHRR and Landsat TM data.Journal of directions for occurrence of fire were the east, north-east Geophysical Research, 102(24): 29581-29598.and west.There was a significant difference in fire 9. Kemp, D.D., 1990.Global Environmental Issues.intensities (confidence levels) between forest plantations A Climatological Approach, Routledge.and cultivation as well as between forest plantations and 10.Albright, D. and B.N. Meisner,1999.Classification of grazing.The fire intensity was higher for the forest Fire Simulation Systems.Fire Management Notes, plantations compared to the cultivated land and grazing 59(2): 45-63.land.About 29% of the fires were causes by honey 11.Justice, C., L. Giglio, S. Korontzi, J. Owens, hunters and this was prevalent in forest plantations.Out J. Morisette and D. Roy, 2002.The MODIS fire of the 99 wildfires, 51 were not suppressed and they products.Remote Sensing and Environment, caused the maximum possible destruction.The forest 83(2): 244-262.plantations had their own trained fire brigades yet 12. Smith, A.M.S., N.A. Drake, M.J. Wooster, communities lacked training in fire management and A.T. Hudak, Z.A. Holden and C.J. Gibbons, 2007.suppression.There were no early warning systems or Production of Landsat ETM+ reference imagery of alarms to inform rural communities about occurrence of burned areas within Southern African savannahs: fire.The communities lacked incentives to manage fires, Comparison of methods and application to MODIS, especial when the fires were in adjacent private farms.
7. Burrows, N.D., B. Ward and A. Robinson, 2009. of western desert.Proceedings of the Royal Society