An Analysis of the Causes of Deforestation in Malawi: A Case of Mwazisi

: Deforestation is recognized as a major driver of the loss of biodiversity and ecosystem services. It also disturbs natural processes such as biogeochemical, hydrological, and ecological cycles. In Malawi, deforestation is estimated to be responsible for the loss of 33,000 hectares per year, and is mainly attributed to agriculture expansion, tobacco growing, and excessive use of biomass. However, little research has been conducted at either the local level or that of forests located on customary land. This research aimed to identify and analyze the underlying driving factors associated with the proximate factors of agriculture expansion, tobacco growing, and brick burning in Mwazisi. Landsat images for 1991, 2004, and 2017 were downloaded from the United States Geological Survey website and used to analyze changes in forest cover. Interviews with households ( n = 399) and Natural Resource Committee members, a focus group discussion with key ofﬁcers, and observations were conducted during ﬁeld data collection in 2017. The results of the land cover analysis showed that forest covered 66% of the study area in 1991, and by 2017 it had decreased to 45.8%. Most households depend on wood from customary land forests for tobacco curing (69%) and brick burning (68%). Furthermore, 47.6% of the households have expanded their agriculture land by approximately 0.57 hectares during the past 15 years. The interview survey and the focus group discussion identiﬁed that the underlying driving factors towards these anthropogenic activities are: (a) population growth, (b) poverty, (c) expensive alternative building materials, (d) lack of awareness, (e) lack of resources, (f) lack of commitment from the tobacco companies, and (g) market system of the cash crops grown in the area. In conclusion, a set of economic, institutional, social, and demographic factors, which are associated with imbalanced relationship between rural and urban areas, underpin agriculture expansion, tobacco growing, and brick burning, and have thereby contributed to the decline of the forest cover in Mwazisi, Malawi. comprising forests and shrubs, and (2) non-forest cover, comprising built-up, grass, and agriculture lands. Roofing materials, such as grass, made it difficult to separate built-up areas from either agriculture or grass land. Before spatial analysis and a temporal comparison, accuracy assessment was performed on each image by randomly selecting 100 sample points. The sample points were then imported into the Google Earth for comparison. The overall accuracy of classification was 96% for 1991 image, 94% for 2004 image, and 93% for 2017 image. The Kappa coefficients of the classification were 0.92, 0.88, and 0.86 for 1991, 2004, and 2017, respectively. Socioeconomic data on proximate and underlying factors [30] were collected using interviews, focus group discussion, and field observations in August 2017; these methods are used by many researchers in analyzing the factors of deforestation (for example, [34,43,44]) . Structured interviews were conducted with 399 (256 females, 141 males) heads of households using random sampling technique—66 households were randomly selected in five VDC areas and 69 in one VDC area.


Introduction
Deforestation is known as one of the most important elements for changes in land use and land cover. It is recognized as a major driver of the loss of biodiversity and ecosystem services. Globally, it has been occurring at an alarming rate of 13 million hectares per year [1]. It is believed that high population growth coupled with the rapid expansion of agriculture is responsible for the accelerated rates of deforestation, especially in developing countries [2].
Malawi is a developing country in which enormous pressure is being exerted on forest resources. Forest cover of the country reduced from 47% in 1975 to 36% in 2005 [3]. This is the highest deforestation rate in the Southern African Development Community (SADC) region, representing a net loss of some 30,000 to 40,000 hectares per year [3]. This forest loss is mainly attributed to agriculture expansion and excessive use of biomass, such as wood, charcoal, and agricultural residues mostly used for cooking the proximate factors of deforestation on customary land in the rural area of Mwazisi, Malawi ( Figure  1), where no research about the drivers of deforestation has been conducted to date.

Study Area
The Mwazisi zone, which is the customary land, is located to the west of the Rumphi district in the northern region of Malawi. It consists of six Village Development Committees (VDCs) under the Traditional Authority Chikulamayembe. Traditional Authority is a form of leadership in which the authority of an organization or a ruling regime is largely tied to tradition or custom. The Mwazisi zone is located along Vwaza Marsh Game Reserve (VMGR) and covers an area of 117 km², which contains 1126 households. The total population of the study area is estimated to be approximately 6570. The area is mostly covered by Miombo woodlands with an average temperature of 22.5 °C in the hot dry season [36]. The highest average annual precipitation falls in the month of January is 191 mm.

Methods
To detect changes in forest cover over a long time span [37][38][39][40][41], Landsat 5 Thematic Mapper (19 June, 1991 and6 June, 2004) and Landsat 8 Operational Land Imager (25 May, 2017) images were downloaded from the United States Geological Survey website [37]. ArcGIS 10.2 software was used to preprocess (radiometric correction) the images. Preprocessing of satellite images prior to change detection is essential and aims to establish a more direct linkage between the data and biophysical phenomena [42]. The forest cover maps were generated using a maximum likelihood classifier. Owing to an overlap of spectral signatures among classes, land cover was classified into two categories: (1) forest cover, comprising forests and shrubs, and (2) non-forest cover, comprising builtup, grass, and agriculture lands. Roofing materials, such as grass, made it difficult to separate builtup areas from either agriculture or grass land. Before spatial analysis and a temporal comparison, accuracy assessment was performed on each image by randomly selecting 100 sample points. The sample points were then imported into the Google Earth for comparison. The overall accuracy of classification was 96% for 1991 image, 94% for 2004 image, and 93% for 2017 image. The Kappa coefficients of the classification were 0.92, 0.88, and 0.86 for 1991, 2004, and 2017, respectively.
Socioeconomic data on proximate and underlying factors [30] were collected using interviews, focus group discussion, and field observations in August 2017; these methods are used by many researchers in analyzing the factors of deforestation (for example, [34,43,44]). Structured interviews were conducted with 399 (256 females, 141 males) heads of households using random sampling technique-66 households were randomly selected in five VDC areas and 69 in one VDC area. The

Study Area
The Mwazisi zone, which is the customary land, is located to the west of the Rumphi district in the northern region of Malawi. It consists of six Village Development Committees (VDCs) under the Traditional Authority Chikulamayembe. Traditional Authority is a form of leadership in which the authority of an organization or a ruling regime is largely tied to tradition or custom. The Mwazisi zone is located along Vwaza Marsh Game Reserve (VMGR) and covers an area of 117 km 2 , which contains 1126 households. The total population of the study area is estimated to be approximately 6570. The area is mostly covered by Miombo woodlands with an average temperature of 22.5 • C in the hot dry season [36]. The highest average annual precipitation falls in the month of January is 191 mm.

Methods
To detect changes in forest cover over a long time span [37][38][39][40][41], Landsat 5 Thematic Mapper (19 June, 1991 and6 June, 2004) and Landsat 8 Operational Land Imager (25 May, 2017) images were downloaded from the United States Geological Survey website [37]. ArcGIS 10.2 software was used to preprocess (radiometric correction) the images. Preprocessing of satellite images prior to change detection is essential and aims to establish a more direct linkage between the data and biophysical phenomena [42]. The forest cover maps were generated using a maximum likelihood classifier. Owing to an overlap of spectral signatures among classes, land cover was classified into two categories: (1) forest cover, comprising forests and shrubs, and (2) non-forest cover, comprising built-up, grass, and agriculture lands. Roofing materials, such as grass, made it difficult to separate built-up areas from either agriculture or grass land. Before spatial analysis and a temporal comparison, accuracy assessment was performed on each image by randomly selecting 100 sample points. The sample points were then imported into the Google Earth for comparison. The overall accuracy of classification was 96% for 1991 image, 94% for 2004 image, and 93% for 2017 image. The Kappa coefficients of the classification were 0.92, 0.88, and 0.86 for 1991, 2004, and 2017, respectively.
Socioeconomic data on proximate and underlying factors [30] were collected using interviews, focus group discussion, and field observations in August 2017; these methods are used by many researchers in analyzing the factors of deforestation (for example, [34,43,44]). Structured interviews were conducted with 399 (256 females, 141 males) heads of households using random sampling technique-66 households were randomly selected in five VDC areas and 69 in one VDC area. The sample size was calculated using Cochran's (1963:75) formula [45] below to secure the representativeness of the community: where n is the sample size, Z is constant, e is the level of precision, and p is the estimated proportion of an attribute. The purpose of the household survey was to understand socioeconomic conditions, forest dependency, and the awareness of local people regarding forest use and management, i.e., to examine economic and institutional factors as the underlying factors. Interviews were also conducted with two officers from the Forestry Department and one officer from the Department of Parks and National Wildlife, as well as two members of the Natural Resources Committee (NRC), to identify problems and the current condition of forests in the study area. The focus group discussion was conducted with officers from the Department of Agriculture regarding forest management and challenges. Through the interviews and the focus group discussion, institutional factors as part of the underlying factors were examined.
Field observations were also conducted in the farms and forests during the social surveys in 2017. Photographs were taken to illustrate some of the causes of deforestation in the study area.
Secondary data such as, the prices of tobacco, maize, groundnuts, and soybeans and tobacco farming practices, were collected from the Tobacco Control Commission of Malawi (TCCM), the Agricultural Development and Marketing Corporation (ADMARC), the National Association of Smallholder Farmers (NASFAM), the Labor Office, the Forestry Department, and the reports to analyze the economic factors.

Forest Change Analysis
The results of the classification show that forest was the dominant land cover in the year 1991 ( Figure 2a); however, it has declined tremendously over the years. Forest covered 66% of the area in 1991 and decreased to 45.8% in 2017 ( Table 1). The annual rate of forest cover loss between 1991 and 2004 was 1.3% and increased to 1.6% in the period between 2004 and 2017.  [45] below to secure the representativeness of the community: where n is the sample size, Z is constant, e is the level of precision, and p is the estimated proportion of an attribute. The purpose of the household survey was to understand socioeconomic conditions, forest dependency, and the awareness of local people regarding forest use and management, i.e., to examine economic and institutional factors as the underlying factors. Interviews were also conducted with two officers from the Forestry Department and one officer from the Department of Parks and National Wildlife, as well as two members of the Natural Resources Committee (NRC), to identify problems and the current condition of forests in the study area. The focus group discussion was conducted with officers from the Department of Agriculture regarding forest management and challenges. Through the interviews and the focus group discussion, institutional factors as part of the underlying factors were examined.
Field observations were also conducted in the farms and forests during the social surveys in 2017. Photographs were taken to illustrate some of the causes of deforestation in the study area.
Secondary data such as, the prices of tobacco, maize, groundnuts, and soybeans and tobacco farming practices, were collected from the Tobacco Control Commission of Malawi (TCCM), the Agricultural Development and Marketing Corporation (ADMARC), the National Association of Smallholder Farmers (NASFAM), the Labor Office, the Forestry Department, and the reports to analyze the economic factors.

Forest Change Analysis
The results of the classification show that forest was the dominant land cover in the year 1991 ( Figure 2a); however, it has declined tremendously over the years. Forest covered 66% of the area in 1991 and decreased to 45.8% in 2017 ( Table 1). The annual rate of forest cover loss between 1991 and 2004 was 1.3% and increased to 1.6% in the period between 2004 and 2017.

Agriculture Expansion
Interviews show that most households (80.7%) depend on agriculture to support the livelihood while only 19.3% earn their living through business and employment. All hou grow a crop of maize as a staple food, and for the past 15 years, 47.6% of the household expanded their maize farm (Figures 3 and 4). On average, each household has expan agriculture land by approximately 0.57 hectares during the past 15 years. Most households expanded their maize farms due to an increase in family size (on average, each household h children) and a lack of farm inputs. The Pearson product-moment correlation coefficient also a positive correlation between the frequency of the agriculture expansion and the number of c in a household (correlation = 0.3764, p = 4.949 × 10⁻⁶).

Agriculture Expansion
Interviews show that most households (80.7%) depend on agriculture to support their daily livelihood while only 19.3% earn their living through business and employment. All households grow a crop of maize as a staple food, and for the past 15 years, 47.6% of the households have expanded their maize farm (Figures 3 and 4). On average, each household has expanded its agriculture land by approximately 0.57 hectares during the past 15 years. Most households (91.2%) expanded their maize farms due to an increase in family size (on average, each household has four children) and a lack of farm inputs. The Pearson product-moment correlation coefficient also shows a positive correlation between the frequency of the agriculture expansion and the number of children in a household (correlation = 0.3764, p = 4.949 × 10 −6 ).  Interviews show that most households (80.7%) depend on agriculture to support their daily livelihood while only 19.3% earn their living through business and employment. All households grow a crop of maize as a staple food, and for the past 15 years, 47.6% of the households have expanded their maize farm (Figures 3 and 4). On average, each household has expanded its agriculture land by approximately 0.57 hectares during the past 15 years. Most households (91.2%) expanded their maize farms due to an increase in family size (on average, each household has four children) and a lack of farm inputs. The Pearson product-moment correlation coefficient also shows a positive correlation between the frequency of the agriculture expansion and the number of children in a household (correlation = 0.3764, p = 4.949 × 10⁻⁶).

Tobacco Growing
Tobacco is the main cash crop in the area and is grown by 45.4% of households, while the remaining households depend on subsistence farming. Of the tobacco farmers, 46.4% expanded their agriculture land by an average of approximately 0.39 hectares per year. These farmers expanded their agriculture land mainly to increase earnings or profit. The type of tobacco grown in the study area is burley, which is air-cured in barns. Figure 5 shows that 69% of the farmers extracted wood from the forest (including ropes and twigs) to construct the barns ( Figure 6).

Tobacco Growing
Tobacco is the main cash crop in the area and is grown by 45.4% of households, while the remaining households depend on subsistence farming. Of the tobacco farmers, 46.4% expanded their agriculture land by an average of approximately 0.39 hectares per year. These farmers expanded their agriculture land mainly to increase earnings or profit. The type of tobacco grown in the study area is burley, which is air-cured in barns. Figure 5 shows that 69% of the farmers extracted wood from the forest (including ropes and twigs) to construct the barns ( Figure 6).

Tobacco Growing
Tobacco is the main cash crop in the area and is grown by 45.4% of households, while the remaining households depend on subsistence farming. Of the tobacco farmers, 46.4% expanded their agriculture land by an average of approximately 0.39 hectares per year. These farmers expanded their agriculture land mainly to increase earnings or profit. The type of tobacco grown in the study area is burley, which is air-cured in barns. Figure 5 shows that 69% of the farmers extracted wood from the forest (including ropes and twigs) to construct the barns ( Figure 6).

Brick Burning
There are three types of building materials in Mwazisi: clay bricks, mud, and wood. Clay bricks are the main building material and are used by 65.7% of the households (Figure 7). Clay bricks are

Tobacco Growing
Tobacco is the main cash crop in the area and is grown by 45.4% of households, while the remaining households depend on subsistence farming. Of the tobacco farmers, 46.4% expanded their agriculture land by an average of approximately 0.39 hectares per year. These farmers expanded their agriculture land mainly to increase earnings or profit. The type of tobacco grown in the study area is burley, which is air-cured in barns. Figure 5 shows that 69% of the farmers extracted wood from the forest (including ropes and twigs) to construct the barns ( Figure 6).

Brick Burning
There are three types of building materials in Mwazisi: clay bricks, mud, and wood. Clay bricks are the main building material and are used by 65.7% of the households (Figure 7). Clay bricks are  There are three types of building materials in Mwazisi: clay bricks, mud, and wood. Clay bricks are the main building material and are used by 65.7% of the households (Figure 7). Clay bricks are burned before their use in construction and the source of energy is wood (Figure 8). Of the brick-walled houses in the area, 68% used wood from the forests and 31% used wood left over after the clearing of land for agriculture. Field results show that each brick-walled house used 4 metric tons of wood, on average. burned before their use in construction and the source of energy is wood (Figure 8). Of the brickwalled houses in the area, 68% used wood from the forests and 31% used wood left over after the clearing of land for agriculture. Field results show that each brick-walled house used 4 metric tons of wood, on average.

Underlying Driving Factors
Economic Factors An analysis of the market systems of various crops grown in the area shows that tobacco has a well-developed market structure designed to reach smallholder farmers in the rural areas [5,[46][47][48][49][50]. The tobacco crops are sold to international companies based in the capital city of Lilongwe. The tobacco growing is practiced as a form of contract farming, which helps smallholder farmers by providing access to the market, inputs, and extension services [51]. That is, tobacco companies provide loans, expertise, and transportation of the farm produce to the tobacco market. However, it is more expensive and difficult for smallholder farmers to obtain expertise and loans on crops such as ground nuts, maize, and soybeans. This has resulted in an increase in the number of tobacco farmers despite its impact on the forests and environment.
A comparison of the average price of tobacco crop with others grown in the area, such as maize, groundnuts, and soybeans, shows that tobacco has had the highest average price over the years (Figure 9a). This motivated 71.3% of the farmers while the availability of loans and the market motivated 28.8%. However, a comparison of the average yield per hectare per year (Figure 9b) shows that maize has the highest average yield, followed by groundnuts.  burned before their use in construction and the source of energy is wood (Figure 8). Of the brickwalled houses in the area, 68% used wood from the forests and 31% used wood left over after the clearing of land for agriculture. Field results show that each brick-walled house used 4 metric tons of wood, on average.

Underlying Driving Factors Economic Factors
An analysis of the market systems of various crops grown in the area shows that tobacco has a well-developed market structure designed to reach smallholder farmers in the rural areas [5,[46][47][48][49][50]. The tobacco crops are sold to international companies based in the capital city of Lilongwe. The tobacco growing is practiced as a form of contract farming, which helps smallholder farmers by providing access to the market, inputs, and extension services [51]. That is, tobacco companies provide loans, expertise, and transportation of the farm produce to the tobacco market. However, it is more expensive and difficult for smallholder farmers to obtain expertise and loans on crops such as ground nuts, maize, and soybeans. This has resulted in an increase in the number of tobacco farmers despite its impact on the forests and environment.
A comparison of the average price of tobacco crop with others grown in the area, such as maize, groundnuts, and soybeans, shows that tobacco has had the highest average price over the years (Figure 9a). This motivated 71.3% of the farmers while the availability of loans and the market motivated 28.8%. However, a comparison of the average yield per hectare per year (Figure 9b) shows that maize has the highest average yield, followed by groundnuts.

Underlying Driving Factors Economic Factors
An analysis of the market systems of various crops grown in the area shows that tobacco has a well-developed market structure designed to reach smallholder farmers in the rural areas [5,[46][47][48][49][50]. The tobacco crops are sold to international companies based in the capital city of Lilongwe. The tobacco growing is practiced as a form of contract farming, which helps smallholder farmers by providing access to the market, inputs, and extension services [51]. That is, tobacco companies provide loans, expertise, and transportation of the farm produce to the tobacco market. However, it is more expensive and difficult for smallholder farmers to obtain expertise and loans on crops such as ground nuts, maize, and soybeans. This has resulted in an increase in the number of tobacco farmers despite its impact on the forests and environment.
A comparison of the average price of tobacco crop with others grown in the area, such as maize, groundnuts, and soybeans, shows that tobacco has had the highest average price over the years  The alternative building materials, such as cement (raw material for cement bricks) are regarded as expensive by most households (96.9%), considering that most of the households live below the national poverty line ( Figure 10). The alternative energy sources for brick burning recognized by households (12%) in the study area are animal manure, crop remains, and petroleum; however, the interview survey indicated that these households also lack technical knowledge.

Demographic Factors
There are no population data for the study area; however, there are data for Rumphi ( Figure 11). There is an increasing trend in population in the district and an annual population growth rate of Rumphi is 3.4% [52]. This has resulted in an increase in demand for land for both settlement and agriculture.    The alternative building materials, such as cement (raw material for cement bricks) are regarded as expensive by most households (96.9%), considering that most of the households live below the national poverty line ( Figure 10). The alternative energy sources for brick burning recognized by households (12%) in the study area are animal manure, crop remains, and petroleum; however, the interview survey indicated that these households also lack technical knowledge.

Demographic Factors
There are no population data for the study area; however, there are data for Rumphi ( Figure 11). There is an increasing trend in population in the district and an annual population growth rate of Rumphi is 3.4% [52]. This has resulted in an increase in demand for land for both settlement and agriculture.  The alternative building materials, such as cement (raw material for cement bricks) are regarded as expensive by most households (96.9%), considering that most of the households live below the national poverty line (Figure 10). The alternative energy sources for brick burning recognized by households (12%) in the study area are animal manure, crop remains, and petroleum; however, the interview survey indicated that these households also lack technical knowledge.

Demographic Factors
There are no population data for the study area; however, there are data for Rumphi ( Figure 11). There is an increasing trend in population in the district and an annual population growth rate of Rumphi is 3.4% [52]. This has resulted in an increase in demand for land for both settlement and agriculture.

Demographic Factors
There are no population data for the study area; however, there are data for Rumphi ( Figure 11). There is an increasing trend in population in the district and an annual population growth rate of Rumphi is 3.4% [52]. This has resulted in an increase in demand for land for both settlement and agriculture.

Institutional Factors
The Forest Act is a fundamental tool for proper forest use and management of private, customary, and public land in Malawi. The results from the field survey, however, show that 95.2% of households are unfamiliar with the Act. Most households (97.7%) are unaware of the prohibition of forest wood extraction for brick burning. Furthermore, 97.8% of tobacco farmers are unaware of the prohibition of forest wood extraction for tobacco processing.
The focus group discussion and interviews with the officers from agriculture and forestry reveal the existence of financial and material constraints in the district. This has led to a reduction in field activities, such as monitoring, awareness campaigns, and law enforcement, especially on customary land forests. With few resources in the district, priority is mostly given to the forest reserves (one gazette and three proposed forest reserves).
Tobacco companies have been involved in deforestation mitigation activities, notably tree planting. However, the initiative has yielded few results. Field survey data show that approximately 10,980 tree seedlings were distributed to tobacco farmers by four tobacco companies in 2016. The quantity of tree seedlings given to each farmer is determined by the size of the farm (i.e., 130 trees seedlings per 0.5 hectares). Almost all farmers (94%) planted the seedlings; however, only approximately 257 seedlings survived. The farmers complained about the late distribution of the seedlings (usually distributed towards the end of the rainy season), which resulted in the low survival of the planted seedlings. The focus group discussion and interviews identified that the four tobacco companies do not monitor their farmers while they are planting and caring for the distributed seedlings. Furthermore, there is lack of collaboration between the tobacco companies and governmental departments. That is, the companies rarely share information, resulting in officers' failure to follow up on any activities conducted by the tobacco companies.

Interaction between Underlying and Proximate Factors of Deforestion
This study revealed the existence of multiple underlying driving factors towards the proximate factors of agriculture expansion, tobacco growing, and brick burning. Furthermore, this study identified that each underlying driving factor underpins one or multiple proximate factors, as shown in Figure 12.
factors of agriculture expansion, tobacco growing, and brick burning. Furthermore, this study identified that each underlying driving factor underpins one or multiple proximate factors, as shown in Figure 12.
An analysis of the satellite images shows a significant areal reduction in the forest cover; it decreased from 66.0% in 1991 to 45.8% in 2017 ( Table 1). The high rate of the deforestation is partly owing to agriculture expansion, the liberalization of tobacco farming after 1995, and brick burning.  An analysis of the satellite images shows a significant areal reduction in the forest cover; it decreased from 66.0% in 1991 to 45.8% in 2017 ( Table 1). The high rate of the deforestation is partly owing to agriculture expansion, the liberalization of tobacco farming after 1995, and brick burning.
Agriculture supports the livelihood of most households in Mwazisi. Subsistence farming mostly involves crops such as maize, groundnuts, and soybeans (ground nuts and soybeans are usually intercropped with maize). A statistical analysis of the field data showed a positive correlation between the frequency of the agriculture expansion and the number of children in a household (correlation = 0.3764, p-value = 4.949 × 10 −6 , significance value = 0.05). This finding is in line with the literature, which states that the largest net loss of forest area and large gain in agriculture area (in the low income group of countries) are associated with an increase in the rural population [53] as well as urban population. Nonetheless, agriculture expansion depends not only on an increase in the number of children in a household but also on other factors, such as poverty. Households cannot afford sufficient farm inputs, such as fertilizer, to restore soil fertility, which has further resulted in the expansion of cultivation land to increase the harvest yield. The poverty ratio is higher in rural areas than in urban areas in Malawi-about 43% of the population in the rural areas reside in poverty, compared with 14% of the urban population [54]. Most smallholder farmers in the rural areas of Malawi may have little access to fertilizer because of its high price [55].
With regards to the tree planting initiative in tobacco farming, farmers complained about the late distribution of seedlings which affected the survival of the planted seedlings. A study by Clarkson [56] showed that many farmers would devote their resources to planting trees if they were able to source seedlings at an appropriate time in the year. Appropriate timing for seed distribution and monitoring would help to increase the survival rate of the planted tree seedlings as the interviewees accounted for. Also, the tobacco companies demonstrate a lack of commitment. The tobacco companies do not monitor farmers when they are planting and managing the care of the distributed tree seedlings. In 2016, among approximately 10,980 tree seedlings distributed, 257 seedlings were alive, as described earlier.
These findings are similar to the results obtained by the Extension Service of Malawi, which found that 80% of the estate farmers had failed to follow the government's recommendation to plant trees on 10% of the farm land [11]. The tobacco companies are much more interested in the economic benefits of tobacco and pay little attention to the forest degradation caused by tobacco [11]. Currently, tobacco farmers meet their wood needs with trees from the forests nearby-burley tobacco requires 158 trees for every 0.15 hectares of tobacco [57].
Bricks are the main building materials in both urban and rural settings in Malawi, and the wood needed for brick production is extracted from natural woodlands or forests. The alternative building materials, such as cement, are regarded as expensive, especially for rural residents, because most households live below the national poverty line (Figure 10). When comparing the clay bricks produced in an open kiln with those produced by an alternative method called Vertical Shaft Brick Kiln (VSBK or Eco-kiln), which consumes less fuel and also uses carbonaceous waste or coal, the price of each VSBK brick is three times more expensive than that of bricks produced in an open kiln [58]. Consequently, people opt for the cheaper bricks. The VSBK method also requires a huge amount of capital for its establishment and it mostly targets the urban population, which only constitutes 14.39% of households, compared to households in the rural areas of the northern region, which comprise 86.61% [52,58].

Deforestation on Customary Land
The majority of users of wood energy are found in the customary land in rural areas, where almost 90% of the population lives [52]. According to the literature [59] over 50% of the wood energy in Malawi comes from customary forests and woodlands. Forests on customary land are managed by the rural community; therefore, proper knowledge, support, and empowerment are required, although they are imbalanced between the rural and urban areas. According to Sillah [60], the awareness level of the local population concerning conservation and rational utilization of forest resources must be augmented to acquire the active participation and commitment of communities and individuals. The findings of this study, however, show a low level of awareness among those in the local population regarding forest use and management. The lack of resources at the district level has partly contributed to the problem. For example, the forestry budget for one year (2016-2017) for Rumphi was 9366.87 US$ with a monthly budget of 780.57 US$. This has resulted in a reduction in law enforcement, awareness campaigns, and monitoring, especially for customary land forests as the interviewees accounted for. Developing countries barely meet the financial, material, and personnel requirements for sustainable forest management. People continue to illegally extract wood from customary land forests for either commercial or non-commercial purposes.

Measures to Mitigate Pressure on Forests
Tobacco is an important cash crop in Malawi, as it accounts for 35% of the Gross Domestic Product [3]. The results of this study suggest the existence of a number of factors that motivate farmers to grow tobacco over other cash crops, which include: (1) the availability of loans facilitated by the tobacco companies, (2) a better price for tobacco compared to that of other cash crops, and (3) easy access to tobacco information and the availability of a market for the crop. These results are similar to those of the research conducted by the Centre for Agricultural Research and Development [61]. Hall [53] reported that most governments send out mixed messages regarding their concern for people and the environment, while actively and assiduously promoting the very economic sectors that drive deforestation. If the supply chains for alternative crops were developed to the level of tobacco supply chains, the prices and profitability of these crops would also grow and eclipse those of contract tobacco [56]. This, in turn, would help to reduce the pressure on forest resources exerted by tobacco farming.
The alternative for brick burning in the study area would be an introduction of stabilized soil bricks (SSBs) [16,62] and promotion of its use. This method involves the use of either soil alone or a mixture of soil and a minimum amount of 10% cement. The mixed soil and cement are compressed at high pressure and are cured under a shade [16]. This method produces bricks using very little or no energy; therefore, this alternative would lead to reduction of deforestation.

Limitation of the Study
Although this study found a decrease in forest cover, higher spatial resolution images would be able to separate overlapping classes, such as forest and shrubs, and built-up grass and agriculture lands, more accurately. The results of the social survey may be applied to some local areas in the country, but an accumulation of similar studies is necessary to understand the similarities and differences among its local levels.

Conclusions
Landsat images were used to assess forest cover changes of the study area. Forest cover in Mwazisi was reduced from 66% in 1991 to 45.8% in 2017. Qualitative and quantitative methods were used to assess socioeconomic conditions, forest dependency, and the underlying driving factors of deforestation. Households continue to depend on forest resources for (1) agriculture expansion, (2) tobacco curing, and (3) brick burning. The underlying factors towards these anthropogenic factors are the market system, poverty, and population growth, expensive alternative building materials, lack of awareness, lack of resources, and lack of commitment. Each of these underlying drivers of deforestation interacts with single or multiple proximate factors. Additionally, there are multiple underlying driving factors working together to underpin each proximate factor of deforestation, thereby impacting the forest cover reduction in Mwazisi. Synergies also exist between some underlying driving factors, such as a lack of awareness and resources. A set of economic, institutional, and demographic factors underpin agriculture expansion, tobacco growing, and brick burning in Mwazisi, Malawi. The following recommendations would facilitate the reduction in the deforestation rate: Providing technical support to the village heads and Community-Based Natural Resources Management Committee on forest management, and monitoring the tobacco companies operating in the district.