Spatial and Temporal Dynamics of Land Use and Land Cover In and Around Magamba Nature Reserve

TABLE OF CONTENTS

ABSTRACT

COPYRIGHT

DEDICATION

LIST OF TABLES

LIST OF FIGURES

LIST OF APPENDICES

LIST OF ABBREVIATIONS

CHAPTER ONE 1.0 INTRODUCTION
1.1 Background Information
1.2 Problem Statement and Justification
1.3 Objectives of the Study
1.3.1 General objective
1.3.2 Specific objectives
1.4 Research Questions`
1.5 Hypothesis

CHAPTER TWO 2.0 LITERATURE REVIEW
2.1 Land Use and Land Cover
2.2 Drivers of Land use and Land Cover Changes
2.3 Human Population and Land Use Change
2.4 Application of Remote Sensing and Geographic Information Systems for Land Use and Land Cover Changes Analysis
2.5 Concept of Land Use and Land Cover Classification
2.6 Remote Sensing in Land Use and Land Cover Change Detection
2.7 Impacts of Land Use and Land Cover Changes on Nature Reserves

CHAPTER THREE 3.0 MATERIALS AND METHODS
3.1 Description of the Study Area
3.1.1 Location
3.1.2 Geology and soils
3.1.3 Climate
3.1.4 Natural vegetation
3.1.5 Population and ethnicity
3.1.6 Economic activities
3.2 Data Description and Collection
3.2.1 Spatial data collection
3.2.2 Population data
3.2.3 Topographic maps and google/aerial photographs
3.2.4 Identification of socio-economic factors influencing land use and land cover changes
3.2.4.1 Sampling procedure
3.2.4.2 Participatory rural appraisal
3.2.4.3 Questionnaire administration
3.2.4.4 Participant observations
3.2.4.5 Focused group discussion
3.3 Data Analysis
3.3.1 Graphical illustration of the methodology
3.3.2 Image processing and analysis
3.3.3 Image pre-processing
3.3.4 Image rectification/Geo-referencing
3.3.5 Image enhancement
3.3.6 Image classification
3.3.6.1 Supervised classification
3.3.6.2 Descriptions of land use and land cover categories for the study area
3.3.7 Ground truthing
3.3.8 Accuracy assessment
3.3.8.1 Error matrix of 1995 classified image
3.3.9 Land use and land cover change detection analysis
3.3.10 Assessment of the rate of cover change
3.4 Land Use and Land Cover-Population Relationship Analysis
3.5 Social Economic Data Processing and Analysis
3.5.1 Data analysis
3.5.1.1 Qualitative data analysis
3.5.1.2 Quantitative data analysis

CHAPTER FOUR 4.0 RESULTS
4.1 Land Use and Land Cover Changes during the Period 1995-
4.1.1 Cover area, changed area and the rate of change between 1995-
4.1.2 Cover area, changed area and the rate of change between 2008 -
4.1.3 Change transitions in land use and land cover between 1995 and
4.1.4 Quantifying the “from – to” change
4.1.4.1 Change detection matrix of land use and land cover types for the period 1995 –
4.1.4.2 Change detection matrix of land use and land cover types for the period 2008 –
4.2 Socioeconomic Characteristic of Respondents
4.2.1 Gender of respondents
4.2.2 Age of respondents
4.2.3 Marital status of respondents
4.2.4 Education level
4.2.5 Main household’s income generating activities
4.2.6 Household land size and land acquisition
4.2.7 Households energy sources
4.2.8 Local community’s perception on the importance of Magamba nature reserve
4.2.9 Perception of local communities on drivers of land use and land cover change
4.3 Population Growth
4.3.1 Population densities of the villages around Magamba nature reserve
4.3.1.1 Population density for the period 1995 –
4.3.1.2 Population density for the period 2008 –
4.3.1.3 Population density for the period 1995 –
4.3.1.4 Population growth and land use and land cover patterns

CHAPTER FIVE 5.0 DISCUSSION
5.1 Land Use and Land Cover Dynamics
5.1.1 Accuracy assessment
5.2 Socioeconomic Conditions of the Respondents
5.2.1 Gender of respondents
5.2.2 Age of respondents
5.2.3 Education level
5.2.4 Main household’s income generating activities
5.2.5 Household land size and land acquisition
5.2.6 Households energy sources
5.3 Perception of Local People on Causes of Land Use and Land Cover Changes
5.4 Population Growth

CHAPTER SIX 6.0 CONCLUSIONS AND RECOMMENDATIONS
6.1 Conclusions
6.2 Recommendations

REFERENCES

APPENDICES

LIST OF TABLES

Table 1: Landsat images used in the analysis of land use and land cover changes

Table 2: Error matrix representing accuracy of supervised classification for the image scene of 1995

Table 3: Accuracy totals and Kappa Statistics result for 1995 classified image

Table 4: Error matrix representing accuracy of supervised classification for the image scene of 2008

Table 5: Accuracy totals and Kappa Statistics result for 2008 classified image

Table 6: Error matrix representing accuracy of supervised classification for the image scene of 2015

Table 7: Accuracy totals and Kappa Statistics result for 2015 classified image

Table 8: Areas and extents of respective land use and land cover class in and around the Magamba Nature Reserve for each study year in Tanzania

Table 9: Cover Area, Changed Area and the Rate of Change between 1995 and 2008 for the study area

Table 10: Cover Area, Changed Area and the Rate of Change between 2008 and 2015 for the study area

Table 11: Change Detection Matrix of Different Land Use and Land Cover for the Period 1995 - 2008

Table 12: Change Detection Matrix of Different Land Use and Land Cover for the Period 2008-2015

Table 13: Gender of the respondents in three villages that are proximity to Magamba Nature Reserve, Tanzania

Table 14: Age of respondents in three villages that are proximity to Magamba Nature Reserve, Tanzania

Table 15: Marital status of respondents in three villages that are proximity to Magamba Nature Reserve, Tanzania

Table 16: Education level of respondents in three villages that are proximity to Magamba Nature Reserve, Tanzania

Table 17: The main household income generating activities of respondents around Magamba Nature Reserve, Tanzania

Table 18: Land size and acquisition of interviewed households around Magamba

Table 19: Chi-square test for socioeconomic factors influencing land use and l and cover changes

Table 20: Population densities of the villages around Magamba Nature Reserve from 1995 to 2015, Tanzania

LIST OF FIGURES

Figure 1: Location of Magamba Nature Reserve in Lushoto District, Tanzania

Figure 2: Graphical illustration of the methodology

Figure 3: Distribution of ground control points

Figure 4: Distribution in land use and land cover within and around Magamba Nature Reserve between 1995 – 2015

Figure 5: Increase and decrease in land use and land cover within and around Magamba Nature Reserve between 1995 – 2015

Figure 6: Land use and land cover map of Magamba Nature Reserve in 1995

Figure 7: Land use and land cover map of Magamba Nature Reserve in 2008

Figure 8: Land use and land cover map of Magamba Nature Reserve in 2015

Figure 9: Transition from which the land use and land cover of 1995 changed to other types of land use and land cover in 2015

Figure 10: Households energy consumption of communities around Magamba Nature Reserve

Figure 11: Source of energy of interviewed households around Magamba Nature Reserve

Figure 12: Perception of interviewed households on importance of Magamba Nature Reserve

Figure 13: Local communities’ perception on drivers of land use and land cover change in and around Magamba Nature Reserve

Figure 14: Total population of the villages surrounding Magamba Nature Reserve from 1995 to 2015

Figure 15: Population of each village surrounding Magamba Nature Reserve

Figure 16: Population growth in respective village surrounding Magamba Nature Reserve

Figure 17: Population densities per square kilometers of the villages around Magamba Nature Reserve for the period 1995 – 2008

Figure 18: Population densities per square kilometers of the villages around Magamba Nature Reserve for the period 2008 – 2015

Figure 19: Population densities per square kilometers of the villages around Magamba Nature Reserve for the period 1995 – 2015

Figure 20: High population density villages and the land use and land cover changes

LIST OF APPENDICES

Appendix 1: Ground truthing field form for collecting Ground control points

Appendix 2: Ground control points (Gcp’s)

Appendix 3: Population data

Appendix 4: Household questionnaire

Appendix 5: Checklist for key informants

LIST OF ABBREVIATIONS

illustration not visible in this excerpt

ABSTRACT

Magamba Nature Reserve in Lushoto District Tanzania has been and continues to experience major land use changes and land cover loss in natural vegetation. This has resulted in biodiversity loss, local climate change, soil erosion and forest degradation. Therefore, understanding of changes occurring in such ecosystem is a central for establishing management activities. It has been lately indicated that these problems are mainly due to anthropogenic activities as this district is among the rural areas with high population growth compared to other rural districts in Tanzania. However, these insights are little proved quantitatively. This study therefore assessed land use and land cover changes and the causes of such changes in and around Magamba Nature Reserve. Remote sensing and GIS techniques were used to quantify and analyze the trend in land use land cover changes over the past 20 years whereby satellite images of 1995, 2008 and 2015 were used. Moreover, household surveys, direct field observations and focus group discussions were employed to obtain socioeconomic factors that influence changes in land use and land cover. Population Census data of 2002 and 2012 were utilized to identify population density for the villages surrounding the reserve. The results indicated a major expansion of agriculture from 169.33 ha in 1995, 282.16 ha in 2008 to 902.54 ha in 2015. There was an increase of built-up areas from 36.50 ha in 1995, 911.56 ha in 2008 to 1792.92 ha in 2015 at the expense of other land covers. The identified influences for such changes include population pressure, unemployment, landless, agricultural expansion, fire and need for energy sources (e.g. firewood and charcoal). Recent increase in population and anthropogenic activities is a threat to conservation and thus should be discouraged in order to restore the degraded areas and for the sustainability of the biodiversity in and around Magamba Nature Reserve.

COPYRIGHT

No part of this dissertation may be reproduced, stored in any retrieval system, or transmitted in any form or by any means without prior written permission of the author or Sokoine University of Agriculture in that behalf.

ACKNOWLEDGEMENTS

First and foremost, I wish to thank my GOD, the Almighty for his blessings throughout the period of the study at Sokoine University of Agriculture. He made it possible to produce this dissertation.

I would like to express my sincere gratitude to my supervisor and advisor Dr. Samora A. Macrice for his scientific mentoring, continuous support, patience, motivation, enthusiasm and immense knowledge. He guided me in all facets of my research and gave me an opportunity to grow academically.

I owe the greatest debt to my adored parents, Mr. Jeremia S. Sembosi and Mrs. Nafikahedi S. Sembosi for their financial support on my studies to this point, love, unremitting encouragement, indispensable prayers and adulation throughout my quest to accomplish this study. They have demonstrated extraordinary courage and made difficult sacrifices. Nothing can fully articulate their roles in making this work a reality. To my brother, Richard Ngoilalei, my uncles and their families: Mr. Mbazingwa Mtaita and Mr. Steven A. Mmasa for valuing my educational life at Sokoine University of Agriculture.

It is not easy to mention all individuals including brothers, sisters, friends and classmates who have contributed to the success of this work. The few mentioned individuals will stand as a tower for all people who have been generous and supportive to see this study a success.

To all, I say be blessed.

DEDICATION

To my wonderful parents, Mr. Jeremia S. Sembosi and Mrs. Nafikahedi S. Sembosi for their love, encouragements, support and prayers were all my strength paved the way for me to recognize the value of education. Also, to my beloved young brothers Abduel J. Sembosi and Abraham J. Sembosi for being the source of encouragement during the period of my study at Sokoine University of Agriculture. May our Almighty God bless them forever.

CHAPTER ONE

1.0 INTRODUCTION

1.1 Background Information

African countries, particularly in the Sub-Saharan are characterized by having most of the individuals living in the rural areas. Their livelihood depends mostly on the utilization of natural resources particularly from the forest, woodlands, wetlands, agricultural land for crops production and livestock keeping (Adger, 2007; Majule, 2013). This dependency coupled with anthropogenic activities through urbanization, agriculture and forestry have been going on and intensified during the past millennium (Soka and Nzunda, 2014). This has altered the earth’s surface significantly and is associated with profound effect upon the natural environment. Moreover, it has resulted in an observable pattern of change in the context of land use and land cover over time.

Among of the impacts caused by changes in the land use and land cover is the alteration in the availability of diverse biophysical resources such as soil, vegetation, water and animal feed (Ohri and Poonam, 2012). Consequently, this leads to a decreased availability of different products and services. This implies that changes on the land are inevitable as far as human beings are in need of fulfilling their basic needs. Furthermore, rapid population increase and subsequent demand for agricultural land and forest harvests such as fuel, forage, timber and lumber had accelerated deforestation practice in numerous developing tropical countries and also producing changes that impact humans (Agarwal et al., 2002; Lambin et al., 2003; Pielke, 2005; Ohri and Poonam, 2012).

A study by Nzunda (2013) in Miombo woodlands documents that changes are powered by a growing demand for agricultural products that are essential for improving food security as well as income. Likewise, in developing world individuals have amplified agricultural harvests mainly by converting more land into agricultural production (Lambin et al., 2003). As Alex (2002) acknowledges that, human production demands cannot be fulfilled without reformation of land covers. Furthermore, poor management of land-based resources, tied with a growing interest and reliance on various products and services poses a challenge for managing the natural resources.

In Tanzania, there have been different cases of encroachments within the protected areas such as forest reserves and national parks. Besides that, conversion of land is a problem such as the transformation of wetlands into agricultural fields. For example, in the highland areas where farmers face the problem of land scarcity and thus expand their land use because of the small land open for expansion of their farms. A witness of this can be made from a study by Misana et al. (2003) who reported a significant expansion of cultivation in Moshi area Tanzania. Additionally, Madoffe et al. (2006) detailed that most of the local communities living around the Eastern Arc Mountains forests depend on the forests for their livelihood. As a result, it has resulted into loss of forest cover in most areas as a result of land clearing for farming activities, unlawful logging, and fire. In addition, changes in land use and land cover conditions could be responsible for the problems associated with hydrological resources in different parts of Tanzania. Therefore, producing more food under conditions of growing water shortage and without creating further environmental degradation is a challenge (CA, 2007).

Integrated management systems, crop water productivity improvement and building the capacity of the people in the natural resources is of great significance. But this needs knowledge on how natural resources evolve over time so as to devise alternative and appropriate strategies for exploiting the resources available. Therefore, it is on the basis of this background that this study was carried out to present the spatial and temporal dynamics on land use and land cover in and around Magamba Nature Reserve (MNR), as one of the forests in highland areas of Tanzania. The study findings will further assist in developing land use planning and good strategies in the management of MNR along with other scarce resources in different places that Tanzania is endowed with.

1.2 Problem Statement and Justification

The natural landscapes have decreased during the past millennium due to human modifications on land (Vanacker, 2005; Nzunda, 2011). Among many factors, population increase has influenced changes on land. This is due to urbanization and mostly land scarcity which has forced farming communities to expand their agricultural fields onto natural forests (Ohri and Poonam, 2012; MNRT, 2015). Moreover, current global agricultural production has already stepped outside the safe boundary and the loss of biodiversity is already outside the upper boundary (Elmqvist et al., 2011; UNFCCC 2011; Nkonya et al., 2012). Apart from all that current understanding of historical land use and land cover change in most parts of Tanzania the situation is not adequate and not well documented as well (Nzunda, 2011; Soka and Nzunda, 2014). Despite the recognition of the magnitude and impact of worldwide changes in land use and land cover, there have been relatively few comprehensive studies on land use and land cover changes and their impacts (Drummond et al., 2010).

Magamba Nature Reserve in West Usambara Mountains has been one of the major locations in Eastern Arc Mountain experiencing changes in land use and land cover due to anthropogenic activities. Some of the consequences of these changes include soil degradation, biodiversity loss and weather changes (Stuart, 1983; Lovett and Stuart, 2001; Collins et al., 2006).

Despite, there is such awareness concerning impacts of human activities on land resources, little quantification has been done within and around the study area. Thus, information is still inadequate as land use and land cover particularly forest ecosystems are still declining over time. Even though increase in population density is perceived to influence changes on Magamba Nature Reserve, these intuitions have not yet been proved quantitatively. Nevertheless, it is not clearly if the perceived changes in this endemic ecosystem are due to increase in population. Therefore, this study entailed to fill this gap by incorporating population to land use and land cover changes. Moreover, to identify socioeconomic factors that alter land use and land cover in and around the study area. The information gathered in this study is not only worthwhile for establishing management baseline data but also for sustainable planning measures to protect natural resources in Magamba Nature Reserve.

1.3 Objectives of the Study

1.3.1 General objective

The overall objective of this study was to assess land use and land cover changes and the causes of such changes in and around Magamba Nature Reserve, Lushoto Tanzania.

1.3.2 Specific objectives

i. To map the trends in land use and land cover in and around the reserve using Landsat images from the period of 1995, 2008 and 2015
ii. To assess the community perception on the socioeconomic factors influencing land use and land cover dynamics in and around the reserve
iii. To establish the spatial relationship between population density and land use and land cover changes

1.4 Research Questions `

i. What the trends in land use and land cover classes in and around the reserve?
ii. To what extent is community aware on the socioeconomic factors that have driven the changes use and land cover from the period of 1995, 2008 and 2015?
iii. What is the relationship between population density and land use and land cover changes?

1.5 Hypothesis

The study is guided by the surmise that:

H0: There is no significant relationship between socioeconomic factors and land use/land cover changes.

H1: There is a significant relationship between socioeconomic factors and land use/land cover changes.

CHAPTER TWO

2.0 LITERATURE REVIEW

2.1 Land Use and Land Cover

Land cover refers to the vegetation (natural and/or planted), water, bare rock, sand and similar surface and also man-made construction occur on the earth’s surface. While land use refers to a series of operations on land, carried out by humans, with the target to obtain products and/or benefit through using land resources including soil resources and vegetation resources which are part of the land cover (DeBie et al., 1996). Thus, land use often influences land cover. In this context, change is defined as a modification of the surface component of the landscape and is only considered to occur if the surface has a different appearance when viewed on at least two successive occasions (Lemlem, 2007).

2.2 Drivers of Land use and Land Cover Changes

Land cover changes may occur due to various factors, which may be broadly divided into natural and human-induced or anthropogenic effects. United States Environmental Protection Agency (EPA, 1999), identified and generalize the causes of land use and land cover changes, to be: (1) natural processes, such as climate and atmospheric changes, wildfire, and pest infestation; (2) direct impacts of human actions, such as deforestation and road-building; and (3) indirect effects of human activity, such as water diversion resulting into dropping of the water table. Even though natural processes may also contribute to modifications in land cover, but the major driving force is human induced land uses (Allen and Barnes, 1985). These human-induced causes of land cover change, which are critical and currently increasing in alarming rate; and can be categorized into two broad divisions: proximate and driving causes. The proximate causes are causes which results immediate to land cover change; while driving, causes are causes which drives behind the immediate causes (Morie, 2007).

According to Alex (2002), the five major factors for driving causes are economics, institutions, technology, culture and demographic change.

Cultural factors include attitudes and perceptions such as unconcern for forests due to little morale and frontier mentalities, lack of stewardship values, and disregard for “nature”, profit-orientation of actors, traditional or inherited modes of cultivation or land exploitation, and a commonly expressed sentiment that it is compulsory to clear the land to establish an exclusive claim.

Institutional factors including policies on land use and economic development, transportation, or subsidies for land-based activities, lack of adequate governance structures, land tenure, and property rights issues, issues of open-access resources and squatting by landless farmers are the major driving causes of cover change.

Demographic factors such as a natural increase or immigration is another driving factor. Most of its explanatory power tend to be derived from interlinkages with other causal forces, especially in the full interplay of all five major drivers.

Technological factors in sectors such as wood and agriculture, like technological changes in the forestry sector in the form of chain saws and heavy implements, and in wood processing, agro-technological factors, modification of farming systems through intensification are playing a significant role in cover change (Morie, 2007).

2.3 Human Population and Land Use Change

For sustainable development land and people are the most important natural resources that are mutually interrelated and interdependent (Pandit, 2011). As the time goes on the use of land is widely increasing as well as intensified as for the increase in use over time. This is due to competion for different uses such as settlements, infrastructure, industry, forests, agriculture, services and recreational areas. Therefore, land use pattern is highly subjective to various interventions by the people and has been experiencing changes significantly (Lee et al., 1991; World Bank, 1994; Junge et al., 2009). For this reason, the matter of land use changes is very significant in the context of population increase as far as when the pressure on land intensified, it leads to both extensive and intensive use of land to meet the needs and wants of the people. Hence, fulfilling the resource required by a growing population ultimately requires more or less form of land use changes in order to provide for the expansion of food production. This is achieved through forest cover clearing, to increase production on already cultivated land or to improve the infrastructure needed to support increasing human population (Hunter, 2000; Lambin et al., 2003).

Population growth and agriculture are two prominent forms of human-induced land use change. During the past three centuries, the amount of World’s cultivated land has grown by more than 45%, increasing from 2.65 million km2 to 15 million km2. At the same time, the world’s forest has been shrinking. Deforestation is closely connected to agricultural land use as a consequence of agricultural expansion. A net decline in forest cover of 180 million acres occurred during the 15 years’ interval 1980-1995, although changes in forest cover vary greatly across the states. On the other side changes in land use, land cover, in particular,have got some impacts on ecosystems (Alados et al., 2004).

For example, farming can lead to soil erosion while the increase in the use of chemical inputs can also lower soil quality. On the other side deforestation also escalate soil erosion, in addition to reducing precipitation due to localized climate dynamics, reducing the ability of soil to hold water and increasing the occurrence and severity of floods. Also land use land cover change results in habitat fragmentation and loss in which is the primary cause of current species decline around the globe. It has been suggested that if current rates of forest clearing continue, a quarter of entire species on earth surface could be lost in the next 50 years (Hunter, 2000; Pandit, 2011).

2.4 Application of Remote Sensing and Geographic Information Systems for Land Use and Land Cover Changes Analysis

Remote sensing can be defined as the use of electromagnetic radiation sensor to record images of the environment which can be interpreted to yield useful information (Paul, 1985; Morie, 2007).Also, Kashaigili (2006), defined as the science and art of obtaining information about an object, area or phenomenon through the analysis of data acquired by a device that is not in contact with the object, area or phenomenon in question. Currently, geographic information system and remote sensing are being increasingly used in combination for spatial analysis purposes. Geographic information system applications and databases are used for extraction of relevant information from remote sensing imagery (satellite images), whereas remote sensing data provide periodic pictures of geometric and thematic appearances of terrain objects, improving our ability to detect changes and update GIS databases (Janssen, 1993). Remote sensing and geographic information systems have been widely recognized and applied as powerful; effective tools in detecting the spatial and temporal changes in land use land cover. Remote sensing provides valuable multitemporal data for monitoring different processes and land-use patterns while GIS techniques make possible the analysis and mapping of different patterns around the globe (Morie, 2007).

In order to understand different environmental dynamics, such as natural and human disturbances, ecological processes like succession and recovery from previous disturbances, can be done through analysis of changing land cover patterns. Satellite imagery provides a tremendous source of data for performing structural studies of the earth’s surface. There are different methods and ways of obtaining information’s from the satellite data such as observation of different patterns, such as the number, size, and shape of patches, can indicate a certain land cover or land use type. When fragmentation statistics are compared to thedifferent time interval, they are useful in describing the type of land use and land cover change and also indicating the resulting impact on the surrounding environment. The areas of land cover transformation between images can also be related to landscape characteristics to determine if the change is due to the presence of certain environmental and/or human-induced influences (Reid, 2000).

Therefore, different techniques and methods of using satellite imageries as data sources have been developed and successfully being useful for land use land cover classification. Moreover, for change detection in various environments such as rural, urban, agricultural and forested areas for biodiversity monitoring through direct and indirect approaches that based on individual organisms and over-reliance on environmental parameters as proxies respectively. In this context, the use of remote sensing satellite data for land use land cover change detection and monitoring is widely applied throughout the globe with the aid of technological improvement that provides high-resolution images (Read and Lam, 2002).

Different information is acquired by the remote sensing system using electromagnetic radiation which has different components such as a source, interactions (both earth’s surface and the atmosphere) and sensor. The amount and characteristics of radiation reflected and/or emitted from the Earth’s surface are dependent and different based on the characteristics of the objects on the Earth’s surface. Therefore, different objects on the Earth interact with radiation at different ways and understanding of this interaction is vital on the classification of satellite images (Paul, 1985). Therefore, based on this reflectance variation of Earth surface objects distinguishing and classification land use land cover is possible.

2.5 Concept of Land Use and Land Cover Classification

There are different methods/ways of analyzing satellite images. Moreover, classification of a satellite image can be achieved by either supervised or unsupervised procedure. Whereby supervised approach is done through specification of training areas by the analyst, in which major land cover types are identified manually as a key for electronically classifying the image unfortunately for this method it requires the knowledge of a study area. In contrast, unsupervised method uses automated methods to cluster reflectance values in order to derive a required number of land classes and their associated spectral signatures.

Land use classification is grounded upon the function and the actual purpose for which the land is currently being used. Hence, a land use can be referred as a series of activities undertaken to produce one or more goods or services (Morie, 2007). A particular land use may be in one, or more than one piece of land and numerous land uses may occur on the same piece of land. Inventory of land by such a classification provides a quantitative measure of land in relation to the economic and environmental outcomes/impacts of different human actions and natural events for precise and quantitative analysis and it's highly supported by ground truthing.

Land cover classification is based on the observed biophysical cover of the earth’s surface regardless of its uses including vegetation, construction works, water, ice, bare rocks or sand surface. Such information is obtained through ground surveys or through remote sensing.

2.6 Remote Sensing in Land Use and Land Cover Change Detection

Remote sensing refers to the technique of obtaining information about an object or feature through the analysis of data obtained by a device that is not in contact with the object or feature under investigation (Lillesand and Kiefer, 2000). Resource managers and planners need a reliable mechanism to assess the consequence of the changes resulted in the stress imposed on the natural resource by detecting, monitoring and analyzing land use changes quickly and efficiently. The conventional technique of environmental data collection and analysis are not efficient in delivering the necessary information in an appropriate and cost effectively approach. Hence viewing the earth from space has become essential to figure out the cumulative influence of human activities on its natural resource base. Remote sensing technology, however, can play a vital role in providing accurate and reliable information with cost effective and lesser time compared to other methods.

Remote sensing delivers a viable source of data from which updated land cover information can be obtained cheaply and efficiently in order to inventory and monitor these changes effectively. Therefore, change detection has turn out to be a major application of remotely sensed data because of repetitive coverage at short intervals and consistent image quality (Mas, 1999).

2.7 Impacts of Land Use and Land Cover Changes on Nature Reserves

Although nature reserves differ from one another, the main purpose of establishing a nature reserve is a way to protect rare, outstanding, or representative natural features or phenomena, such as old-growth forests or the habitats of rare and/or endangered plants or animals. These features can easily be damaged or destroyed by certain types of human activities including forestry, mining, road construction or all-terrain vehicle use.

A nature reserve can be defined as the area of land which is legally protected or designated, under differentacts, laws and/or policies of a particular country, area or state. These areas are protected to safeguard the species, ecosystems, and other natural features, while on the other side providing opportunities for scientific research, education, and nature appreciation (Peterson, 1974). Also,nature reserve can be defined as a protected area of importance for wildlife, flora, fauna, geological features or other special interest, which is reserved and managed for conservation and to provide special opportunities for study or research as well as tourism activities. Moreover, nature reserves may be designated by government institutions in some countries, or by private landowners, such as charities and research institutions, regardless of nationality.

Changes in land use and land cover due to human activities impacts nature reserves in different ways such as habitat destruction, overexploitation of species, introduction of exotic species, pollution and climate changes in both micro and macro environments also global warming. These impact leads to loss of biodiversity which is a growing trend in almost all ecosystems around the globe. According to Millennium Ecosystem Assessment, decline in biodiversity and the related changes in the environment have been more rapid in the past 50 years (MEA, 2005). Therefore, both flora and fauna populations have declined in numbers and/or geographical distribution. However, species extinction occurs naturally within the ecosystems, the current losses mostly are the outcomes of changes in land use land cover as a result of human activities. Based on the report by MEA (2005) anthropogenic activities have increased the extinction rate 100 times more than natural rate (Hansen and DeFries, 2007; Matteucci and Camino 2012; Rija et al., 2013).

CHAPTER THREE

3.0 MATERIALS AND METHODS

3.1 Description of the Study Area

3.1.1 Location

Magamba Nature Reserve (MNR) is located in Lushoto District between latitude 4°40' - 4°46'S and longitude 38°15' - 38°20'E. The district is located in West Usambara Mountains as a part of the Eastern Arc Mountains located in north-eastern Tanzania (4o 24' – 5o 00' S and 38o 10' – 38o 36' E) as indicated in Figure 1. The nature reserve is within the elevation ranging from 900 – 2250 m.a.s.l. The mountains cover an area of 4500km2, which is 90% of the total area of Lushoto district (Msuya and Kideghesho, 2009).

illustration not visible in this excerpt

Figure 1: Location of Magamba Nature Reserve in Lushoto District, Tanzania. “Author’s own work”

3.1.2 Geology and soils

Geologically, Magamba Nature Reserve lies within the basement blocks of the Usambaras whereby they are believed to have been comparatively stable for more than 20 million years and they consist of complex series of ancient metamorphic rocks (Wiersum, 1982). Texturally the rock types may be described as grannies and are often intruded by quartzite veins. Much repetition of outcrop occurs due to complex fold movements although the rock sequence tends to be rather uniform.

The soils are predominantly loams with inconsistent amounts of sand and color that vary from red through gray-brown to black with a pH value ranging between 3.5 and 8.5. They are rich in minerals like Iron, Magnesium, and Manganese. These soils are resistant to erosion under moderate conditions, but if heavily worked under open cultivation they become highly erodible (MNRT, 2008).

3.1.3 Climate

The area has got two rain seasons, which are short rains (November-December) and long rains (March-May), with a rainfall between 600 and 1200 mm per annum. Temperatures are higher in the lower parts (25-27oC mean monthly) and 13-18oC mean monthly at higher levels of the plateau. The minimum and maximum temperatures are 13oC and 27oC, respectively. Lower temperatures of about 7oC during cold seasons and 30oC during hot seasons have been noted (Msuya and Kideghesho, 2009).

3.1.4 Natural vegetation

Magamba Nature Reserve has a diverse of vegetation ranging from lowland, intermediate (sub-montane) and highland (montane) evergreen forests. Generally, Magamba has two main vegetation types with varieties of plant and animal species. The two vegetation types are montane rain and montane dry forests. Wet montane forests are dominated by camphor (Ocotea usambarensis), with some podo (Podocarpus usambarensis and Podocarpus pensiculy), Polyscias fulva, Newtonia buchananii, Entendofragma utirii and have dense undergrowth of Lansthus cirumilee and other shrubs. Associated species include Ficalhoa, Pygium, Rapanea, Fagaropsis and Cassipourea. Dry montane forest consists mainly of cedar (Juniperus procera), with a thick shrub understory of Fuclea, Teclea and Catha species. Other plants include a wide range of grasses and herbs (Mubyazi, 2007; MNRT, 2008).

3.1.5 Population and ethnicity

According to the Tanzania human population census of 2012, Lushoto district has 492 441 inhabitants with an annual growth rate of 1.8% and density of 120/km2 (310/sq. mi). The main ethnic groups are Sambaa, Pare and Mbugu. The Sambaa is the predominant tribe (78% of the population) followed by the Pare (16%). The Mbugu (5%) is the minority. Other small groups are also originating in this area accounting for 1% of the population. The Pare, Mbugu and other tribes are the immigrants to the study area.

3.1.6 Economic activities

Among the economic activities conducted around the area, agriculture is the main income generating activity for most of the individuals even though they practice subsistence farming for their livelihood. The food crops grown are maize, beans, wheat, bananas, Irish potatoes, yams, and cassava, whereas cash crops include tea, coffee, cardamom, sugarcane, fruits (plums, pears, and apples) and vegetables.

3.2 Data Description and Collection

This study used primary and secondary data/information whereby qualitative data were collected by interviews using questionnaires, focus group discussion (FGD) and direct observation. On other side, spatial data particularly the Landsat images (for the year 1995, 2008 and 2015), topographic maps and google/aerial photographs were gathered. Secondary data were collected from various documents both published and unpublished such as management plans, books, journals, articles, magazines and administrative level census data.

3.2.1 Spatial data collection

The used Landsat satellite images were obtained from United States Geological Survey (USGS) Data Interface from the Global Visualization Viewer (GloVis) at http: glovis.usgs.gov i.e. Landsat Thematic Mapper (TM) which are frequently used at a regional study/scale (Lu et al., 2011) and Operational Land Imager (OLI) satellite images. The path and row for the scene covering the study area were identified as well (Table 1).

In this study, several factors were considered while selecting remotely sensed data. That includes study period, scale of the study area, availability of satellite images of the same season free from cloud cover and identifiable features not affected by seasonality and cost/time interval as social and ecological processes (human-environmental dynamics) operate at different scales (Agarwal et al., 2002; Kuemmerle et al., 2013).

Therefore, images were selected from the dry season since it is the best period to obtain the images which are cloud free or minimum cloud coverage. Although it is very difficult to obtain the images free from clouds as Henry et al. (2011) indicates to be a limitation that inherent land sat images, this study succeeded in obtaining the images that were consistent with the study needs.

In addition to image selection criteria, the time points 1995, 2008, and 2015 from TM 5, and OLI 8 land sat were selected to allow for period gaps to detect land use and land cover changes (LULCC). Also, the time points or years selected were destined to correspond with major socioeconomic and administrative changes for the study area in order to capture enough changes.

In addition, during data collection in the field, focused group discussion identified human activities around the nature reserve started progressively in 1995 and population grows fast from 2008 onwards. Therefore, 1995 was taken as a base year before population growth occurred at a higher rate around the nature reserve and 2008 and 2015 being years of higher population growth in the communities surrounding the nature reserve. Furthermore, Global positioning system (GPS) was used to collect ground control points (as presented in Appendix 2) for actual land use and land cover ground verification.

Table 1: Landsat images used in the analysis of land use and land cover changes

illustration not visible in this excerpt

3.2.2 Population data

Population data were acquired for each village surrounding Magamba Nature Reserve from the nature reserve office, village offices as well as the government website of Tanzania (www.geohive.com) and National Bureau of Statistics Ministry of Finance Dar-es-Salaam. Unfortunately, in Tanzania national census is conducted at different periods. Hereafter the population during the inter-census periods was estimated for each village surrounding the Magamba Nature Reserve after being confirmed by overlaying the nature reserve boundary on the Villages shape file.

Thereafter interpolation/extrapolation was done to fill the periods in which census was not conducted and match the data for the villages. The growth trend calculation was used to fill in a series of missing values by means of series of command in Microsoft Excel (URT, 2003).

Exponential growth estimation formula is shown below;

illustration not visible in this excerpt

Where Pl= population at launch year; Pb= population at base year and n is period of time

illustration not visible in this excerpt

Where Pt= is the population in the target year, and z is the number of years in the projection horizon (URT, 2003; Berakhi, 2013).

3.2.3 Topographic maps and google/aerial photographs

The topographical map was obtained from Magamba Nature Reserve and the latest set of Google and aerial photographs covering the study area was obtained from Google Earth, as they were of great significance in guidance during ground truthing. These data were also used to support classification and during accuracy assessment of the classified images.

3.2.4 Identification of socio-economic factors influencing land use and land cover changes

3.2.4.1 Sampling procedure

A cross-sectional research design was used as it studies the population at a single point in time, and data collection is done once (Kothari, 2004). Therefore, due to a limited time for conducting the study, this design was helpful to save time and enough sample was obtained during data collection. Hence, data was collected at one point in time to explore important information on community economic activities as well as their perceptions of land use and land cover changes, forest condition and it was useful to capture an in-depth understanding of historical resources use pattern in the area. Moreover, purposive sampling was used to select three villages (Kwesimu, Gologolo, and Magamba) based on their vicinity to the nature reserve. In addition, 30 households’ heads from each village were randomly selected from village register as a sampling frame obtained from respective village offices as Bailey, (1994) found that a sample size of 30 from one observation unit is considered adequate as Mbeyale (2007) acknowledge.

3.2.4.2 Participatory rural appraisal

This study used Participatory Rural Appraisal (PRA) techniques including Participant observation, focus group discussions and a questionnaire survey with both closed and open-ended questions. This method allows the investigator to participate in the social reality experienced by the community under observation as explained by Kajembe (1994).

3.2.4.3 Questionnaire administration

The questionnaires were addressed to obtain information on the expansion of agricultural activities in and/or around the nature reserve, demand of forest products by the community, education background, immigration, land tenure, knowledge of land use and land cover changes, farm productions, land ownership as well as managements and government restrictions on the use of Magamba Nature Reserve.

3.2.4.4 Participant observations

Participant observations technique was used to make direct observations of local communities’ agriculture practices and forest management activities in the study area. Therefore, the researcher used this method to tie together information collected by other methods (Kajembe and Luoga, 1996).

3.2.4.5 Focused group discussion

Focused group discussions in each village consisted of people with different economic activities, wealth, education background, gender, and age. In addition, government officers and longtime residents were involved mostly in order to obtain in-depth information regarding land use and land cover history of the study area (Luoga et al., 2005).

3.3 Data Analysis

3.3.1 Graphical illustration of the methodology

Below is the figure demonstrating the methodology followed during this study.

illustration not visible in this excerpt

Figure 2: Graphical illustration of the methodology. “Author’s own work”

3.3.2 Image processing and analysis

There are three stages of image processing which are pre-processing, image rectification/geo-referencing and image enhancement as clarified below.

3.3.3 Image pre-processing

Prior to data analysis, initial processing on the raw data (satellite images) was carried out to correct for any distortion due to the characteristics of the imaging system and imaging conditions. Among of the factors are noises that satellite images may have and in consequence this inherent noise in the images can introduce uncertainty into the change detection and may create the appearance of a spurious change, or hinder the identification of a change. Therefore, before any method of change detection on the land use and land cover were applied, proper images were selected followed by layer stacking within QGIS software to merge all the bands at which each of the images was comprising.

3.3.4 Image rectification/Geo-referencing

Re-projection is among of the image processing task that was done before image analysis. In this process satellite images were transformed to a standard projection of a specific area. Within this stage rectification process of the images was carried with reference to topographical map of scale 1:50 000 of the study area to correct for any distortions brought by during the image acquisition process. Moreover, this process is significant in order to ensure geometric compatibility of the satellite images.

Therefore, the Landsat images acquired were georeferenced in coordinate reference system (CRS) World Geodetic System (WGS84), UTM zone 37 North the images were then re-projected to the national coordinate reference system i.e. UTM Zone 37 South, Spheroid Clarke 1880, Datum Arc 1960 in which Magamba Nature Reserve is located. Three rectified images were extracted from the full scenes using Nature Reserve boundary after it was clipped using raster operations in QGIS from the Tanzania Nature Reserves shape file. Then the images were projected in ArcView GIS 3.2a to the national coordinate reference system.

3.3.5 Image enhancement

In order to aid visual interpretation, the visual appearance of the objects in the image can be improved by image enhancement techniques such as grey level stretching to improve the contrast and spatial filtering for enhancing the edges. Therefore, Image enhancement involved mathematical operations that are applied to digital remote sensing input data to improve the visual appearance of an image for better interpretation following digital image analysis (Lillesand et al., 2008).

Normally, individual bands are composited in a Red, Green and Blue (RGB) combination in order to visualize the data in color. There are many different combinations that can be made, and each has their own advantages and disadvantages. For this study both true and false color composite were used to strengthen the visual interpretation of images, a color composite (3, 2 and 1) Red, Green and Blue (True color composite) were applied first for visual interpretation as they style the features on the image to appear in real as it can be seen on the ground/ earth’s surface. However, Band 7, 4 and 2 (False color composite) were used for classification of land use and land cover. All images processing was carried out using QGIS 2.8.1 software. Additionally, Normalized Difference Vegetation Index (NDVI) was also utilized to enhance the best interpretation as a graphical indicator that can be used to analyze vegetation cover. NDVI is calculated as the ratio of (Near infrared - Red Band) to (Near infrared Band + Red Band): NDVI = (Pnir – Pred) / (Pnir + Pred) and it was used within QGIS using Semi-automatic classification plugin.

3.3.6 Image classification

Apart from having different classification systems, this study used NAFORMA classification system. Campbell (2007) describe this system as a widely used LULC classification system which is a reasonable and enduring classification scheme which allows interpretation of features from remotely sensed images. Moreover, this study used satellite images with a resolution of 30m and according to Anderson et al. (1976) is a more appropriate for general information collection and considered for use with Landsat satellite data. Therefore, for images having a resolution of 20 to 100m it’s a more recommended classification system (Lillesand et al., 2008).

Different methods are available for classification and choosing a method depends on the resolution of the image and availability of classification software, among many aspects (Lu et al., 2011). Image classification is the process of sorting pixels into a finite number of individual classes, or categories of data based on their data file values. Therefore, if a pixel satisfies a certain set of criteria, then the pixel is assigned to the class that corresponds to those criteria .

Different land use and land cover types in an image can be categorized using some image classification algorithms using spectral features, i.e. the brightness and "color" information contained in each pixel. There are two primary types of classification algorithm applied to remotely sensed data which are "unsupervised" and/or “supervised".

3.3.6.1 Supervised classification

In supervised classification, the spectral features of some areas of known land cover types are extracted from the image. These areas are known as the "training areas". Every pixel in the whole image is then classified as belonging to one of the classes depending on how close its spectral features are to the spectral features of the training areas. Each class of land use land cover is referred to as a "theme" and the product of classification is known as a "thematic map".

Apart from all, that a supervised approach was used in this study using semi-automatic classification plugin in QGIS software whereby in supervised classification, known region of interest (ROIs) were picked to describe the spectral attributes of each feature type of interest (Lillesand et al., 2008).

This study applied three classification algorithms which are a minimum distance, spectral angle mapping, and maximum likelihood then three different results were obtained regarding the algorithms used. The maximum-likelihood algorithm was chosen to classify images as the results were much suitable compared to other algorithms as prior knowledge of the study area was available in hand. The method group’s together feature in specified classes based on the likelihood of the spectral signature of each feature to the sample set representing a specified class.

3.3.6.2 Descriptions of land use and land cover categories for the study area

Natural forest: According to Food and Agriculture Organization (FAO) defines “forest” as a portion of land bigger than half a hectare (5000m2) with trees higher than 5 meters and a tree canopy cover of more than 10 %, or with trees that will be able to meet these criteria (FAO, 2010). Moreover, it does not include land that is predominantly under agricultural or urban land use. Therefore, natural forest can also be defined as forests of native tree species, where there are no clearly visible indications of human activities and the ecological processes are not significantly disturbed.

Plantation forest: Are forested areas either established artificially by planting or seeding. The trees generally belong to the same specie (whether native or introduced), have the same age and are regularly spaced. The objective of forest plantations can be either for production of wood and/or non - wood goods (productive forest plantations) or the provision of ecosystem services (protective forest plantations).

Grassland: Are areas where the vegetation is dominated by grass (Poaceae), however sedge (Cyperaceae) and rush (Juncaceae ) families can also be found. Furthermore, Grasslands may occur naturally or as the result of human activity.

Shrub-land: It can be defined as open or closed stands of shrubs up to 2 m tall (White, 1983). Moreover, Shrub-land may either occur naturally or be the result of human activity. It may be the mature vegetation type in a particular region and remain stable over time, or a transitional community that occurs temporarily as the effect or result of a disturbance, such as fire.

Woodland: This is a low-density forest forming open habitats with plenty of sunlight and limited shade. Besides that, woodlands may support an understory of other vegetation types such as shrubs and herbaceous plants including grasses.

Agricultural land: Includes a great deal of land not actively or even presently devoted to agricultural use.

Built-up area: Is a term used primarily to reflect a developed area, i.e. any land on which buildings and/or non-building structures are present, normally as part of a larger developed environment

3.3.7 Ground truthing

Different land use/land cover classes were identified from the image scenes of 1995, 2008 and 2015 during classification process. Ground truthing was done in order to verify and modify land covers identified in the initial image interpretation and it was guided by latest google imagery together with a map from Landsat image of 2015. Geographical Position System (GARMIN-GPS device) was used to record ground coordinates for different land use/land cover types as displayed in Figure 3. As far as visual image interpretation is concerned different GIS and RS techniques referred as photographic and geotechnical elements (such as size, texture, shape, tone, drainage, color, association, landform, soil and vegetation) were used to identify and describe different land use/land cover classes (Mascarenhas, 1993; Strindberg and Buckland, 2004). Moreover, local people and nature reserve staffs were very important as they were involved in contributing information's regarding land use/land cover and associated changes within and around Nature Reserve. But also, photos were taken to support ground observations. The recorded coordinates that were used to transform previous land use and land cover types before performing change detection for final results were recorded in the field form (Appendix 2).

illustration not visible in this excerpt

Figure 3: Distribution of ground control points at Magamba Nature Reserve, Tanzania. “Author’s own work”

3.3.8 Accuracy assessment

Due to a number of factors such as classification techniques, data capture and acquisition of satellite images land use land cover maps derived from remote sensing always contain some kind of errors and therefore in order to wisely use the land use land cover maps which are derived from remote sensing and the accompanying land resource statistics, the errors are quantitatively explained in terms of classification accuracy. Depending on the application of the map product whether they meet the expected accuracy or not is usually determined by the users and moreover accuracy levels that are acceptable for the certain task may be unacceptable for others (Mideksa, 2009).

In this study classification accuracy was expressed through the preparation of classification error matrices. Error matrix is a square array of numbers organized in rows and columns which express the number of sample units assigned to a particular category relative to the actual category as indicated by reference data. The table produce many statistical measures of thematic accuracy including user's accuracy, producer's accuracy (User's accuracy represents the probability that a given pixel will appear on the ground as it is classed, while producer's accuracy represents the percentage of a given class that is correctly identified on the map), overall classification accuracy, percentage and the kappa coefficient of agreement which was used to assess the accuracy of the data. Moreover, kappa coefficient indicates how accurate the classification is after accounting for this probability of random chance classification and is calculated by subtracting the change of agreement from the actual agreement of the error matrix (CCRS, 2004). Moreover, Kappa according to Jensen and Cowen (1999) was computed using the equation below;

illustration not visible in this excerpt

Where K is the kappa factor, N is the total number of sites in the matrix, r is the number of rows in the matrix, 𝑥𝑖𝑖 is the number in row i and column i, 𝑥+𝑖 is the total for row i, and 𝑥𝑖+ is the total for column.

Assessment of classification process for the classified images of 1995, 2008 and 2015 was done using error matrix for each year as shown together with accuracy measures. Moreover, for each class, the user’s accuracy, producer’s accuracy, kappa coefficient, overall accuracy and the overall kappa value were computed.

3.3.8.1 Error matrix of 1995 classified image

For the classified image of 1995, the error matrix is shown in Table 2 whereby, natural forest, plantation forest, grassland, shrub-land, woodland, agricultural area and built-up area all showed both commission and omission error but also the majority of the pixels were misclassified to one another and this is due to spectral similarities for the vegetation. For this image scene 40 669-pixel sample units were used for the accuracy assessment.

Table 2: Error matrix representing accuracy of supervised classification for the image scene of 1995

illustration not visible in this excerpt

NF = Natural forest, PF = Plantation forest, GL = Grassland, SL = Shrub-land, WL = Woodland, AL = Agricultural land, BA = Built-up area

Table 3 indicates user’s accuracy, producer’s accuracy, kappa coefficient, overall accuracy and the overall kappa value for the classified image of 1995. The results differ among the land use land cover as indicated in Table 2. Looking at the accuracy assessment the Overall classification accuracy is 95.66 % and overall kappa statistics is 0.93.

Table 3: Accuracy totals and Kappa Statistics result for 1995 classified image

illustration not visible in this excerpt

Overall classification accuracy (%) = 94.30 Overall kappa statistics = 0.91

For the year 2008, the error matrix of the classified image is described in Table 4 whereby 32566-pixel sample units were used for the accuracy assessment. But also, the classified image was having an overall classification accuracy of 96.25 % and overall kappa statistics of 0.92 as Table 5 indicates.

Table 4: Error matrix representing accuracy of supervised classification for the image scene of 2008

illustration not visible in this excerpt

NF = Natural forest, PF = Plantation forest, GL = Grassland, SL = Shrub-land, WL = Woodland, AL = Agricultural land, BA = Built-up area

Table 5: Accuracy totals and Kappa Statistics result for 2008 classified image

illustration not visible in this excerpt

Overall classification accuracy (%) = 96.25 Overall kappa statistics = 0.92

For the classified image of 2015 error matrix are shown in Table 6 and 36518 of the pixel were used as a sample unit during accuracy assessment. Overall classification accuracy of 93.23 % and overall kappa statistics of 0.90 were obtained from this assessment as Table 7 indicates.

Table 6: Error matrix representing accuracy of supervised classification for the image scene of 2015

illustration not visible in this excerpt

NF = Natural forest, PF = Plantation forest, GL = Grassland, SL = Shrub-land, WL = Woodland, AL = Agricultural land, BA = Built-up area

Table 7: Accuracy totals and Kappa Statistics result for 2015 classified image

illustration not visible in this excerpt

Overall classification accuracy (%) = 93.23 Overall kappa statistics = 0.90

3.3.9 Land use and land cover change detection analysis

Once the images were classified for each year, generated land use land cover maps were analyzed following map overlay method for the period 1995-2015. The change detection was done by post classification method using System for Automated Geo-Scientific Analyses (SAGA2.2.0), the method has been found to be the most suitable for detecting land cover changes because it’s the most common approach for comparing data from different sources and dates (Jensen, 1996). Furthermore, the advantage of post-classification comparison is that it bypasses the difficulties associated with the analysis of images acquired at different times of the year and/or by different sensors and this approach identifies changes by comparing independently classified multi-date images on pixel-by-pixel basis using a change detection matrix (Alphan, 2003; Nzunda et al., 2013). The approach identifies quantitative changes by comparing two independent classified images pixel by pixel basis using a change detection matrix (Senga et al., 2014).

3.3.10 Assessment of the rate of cover change

The estimation of the rate of change for the different covers was computed based on the following formulae (Kashaigili, 2006; Kashaigili and Majaliwa, 2010).

illustration not visible in this excerpt

Where: Area i year x = area of cover i at the first date,

Area i year x+1 = area of cover i at the second date,

Abbildung in dieser Leseprobe nicht enthalten= total cover area at the first and

t years = period in years between the first and second scene acquisition dates

3.4 Land Use and Land Cover-Population Relationship Analysis

Population cover relationship was analyzed, based on transition map that was prepared for the periods 1995 to2015 using the MMQGIS plugin in Quantum GIS software (QGIS 2.8.1). The population map was prepared by joining the population data (Appendix 3) to each village shape file that falls in the Magamba Nature Reserve. Therefore, selection by location and overlaying was used in the analysis of population-land use land cover analysis.

3.5 Social Economic Data Processing and Analysis

The data passed through three (3) processing levels that are editing (data collected fresh from the field were assessed to detect errors, omissions, contradictions and unreasonable information to be corrected. This was done to ensure that data is accurate, consistent, uniformly entered and well arranged), coding (an exercise that was carried out after the data collected have been edited, where numerals and other symbols were assigned to questionnaires for easy entry of data in the computer software) and actual analysis, after which presentation of the data in the form of charts, graphs, tables, and figures followed.

3.5.1 Data analysis

3.5.1.1 Qualitative data analysis

Through content and structural-functional analysis qualitative data obtained during focus group discussion and observation were analyzed. Content analysis is a set of methods for analyzing the symbolic content of any communication. The basic idea is to reduce the total content of the communication to some set of categories that represent some characteristics of research interest (Nzunda, 2011; Nzunda, 2013). Therefore, by means of content analysis method, the data collected through verbal discussions were analyzed in details whereby the recorded discussions were broken down into smallest meaningful units of information. According to Kajembe (1994), the technique is useful in explaining the way how social facts relate each other in a social system and the way they relate to the natural physical environment.

3.5.1.2 Quantitative data analysis

After the data, had been edited and coded, then they were entered into the computer for analysis. The actual analysis was done using Statistical Package for Social Science (SPSS) version 16.0 and Microsoft office Excel program. Descriptive and inferential statistics were employed, using the analytical tools such as frequency and chi-square, embedded in the (SPSS), to analyze the socio-economic data. The results of the analyzed data were presented in the form of percentages, frequencies and in forms of tables, charts and graphs.

This study is guided by the decision rule: Reject Ho if p- value is < 0.05 asymp. Sig; otherwise accept Ho and reject H1. This hypothesis was tested using the Pearson chi-squared statistical instrument whose formula was given as (Kothari 2004):

illustration not visible in this excerpt

Whereas:

i = 1, 2, 3 …

j = 1, 2, 3 …

Oij = observed frequencies

Eij = expected frequencies

CHAPTER FOUR

4.0 RESULTS

4.1 Land Use and Land Cover Changes during the Period 1995-2015

The studied area has been identified to have seven land use and land cover classes, which were: natural forest, plantation forest, grassland, shrub-land, woodland, agricultural land, and built-up area. This land use and land cover classes varied between the three time periods under consideration (i.e. 1995, 2008 and 2015).

Considering pattern or distribution of the land use and land cover classes, the classified images reveal significant results verifying the changing aspects observed in terms of the proportion (area) as shown in (Table 8) for the studied periods (1995, 2008 and 2015). During the year 1995 seven land use and land cover were identified. The major land use and land cover was natural forest which accounts for 8051.35 ha (61.06 %), but also 1413.97 ha (10.72 %) for plantation forest, 822.61 ha (6.24 %) for grassland, 358.95 ha (2.72 %) for shrub-land, 2333.37 ha (17.70 %) for woodland, 169.33 ha (1.28 %) for agricultural land and built-up area was 36.50 ha (0.28%). The areas for agriculture and built-up were very small revealing the nature reserve was not that much affected by human activities.

In 2008 the area covered by natural forest, grassland, shrub-land and woodland declined by 6017.27 ha (45.63 %), 1 379.46 ha (10.46 %), 1011.95 ha (7.67 %) and 1 737.77 ha (13.18 %) respectively. On the other side plantation forest increased to 1 845.91 ha (14 %) but also agricultural land together with built-up area emerged with the coverage of 282.16 ha (2.14 %) and 911.56 ha (6.91 %). This implies that human activities within and/or around the nature reserve were active during this period. The results indicate the presence of agricultural lands together with built-up areas which have effects on vegetation cover as Table 8, Figure 4 and 5 indicates a relative decrease of vegetation cover for the period of 13 years (1995-2008).

Table 8: Areas and extents of respective land use and land cover class in and around the Magamba Nature Reserve for each study year in Tanzania

illustration not visible in this excerpt

Classified image of 2015 indicates more decrease in the natural forest such that only 3431.3 ha (26.02 %) remained while the decrease in plantation forest and woodland accounts for 2390.59 ha (18.13 %) and 1216.34 ha (9.22 %). Other land use/land cover accounts for 1670.63 ha (12.67 %) for grassland, 1781.76 ha (13.51 %) for shrub-land, agricultural land 902.54 ha (6.84 %) which decreased compared to the year 2008 and built-up area increased to 1792.92 ha (13.60 %) indicating increase in need of land for settlement for the local communities living around Magamba Nature Reserve due to increased population.

illustration not visible in this excerpt

Figure 4: Distribution in Land Use and Land Cover in and around Magamba

Nature Reserve between 1995 – 2015. “Author’s own work”

illustration not visible in this excerpt

Figure 5: Increase and decrease in Land Use and Land Cover in and around Magamba Nature Reserve between 1995 – 2015. “Author’s own work”

Furthermore, thematic maps were obtained from the satellite images indicating changes occurred for 1995, 2008 and 2015 as presented in Figure 6, 7 and 8 respectively. In general, the maps display the variation in land use land cover between the three-time periods under consideration.

illustration not visible in this excerpt

Figure 6: Land Use and Land Cover Map of Magamba Nature Reserve in 1995. “Author’s own work”

illustration not visible in this excerpt

Figure 7: Land Use and Land Cover Map of Magamba Nature Reserve in 2008. “Author’s own work”

illustration not visible in this excerpt

Figure 8: Land Use and Land Cover Map of Magamba Nature Reserve in 2015. “Author’s own work”

4.1.1 Cover area, changed area and the rate of change between 1995-2008

Regarding changes in cover area and the rate of changes for the period between 1995 to 2015 the results were identified. For 1995 and 2008 the results were that plantation forest, grassland, shrub-land, agricultural land and built-up area increased at a rate of 33.23 ha/year (0.25%/year), 42.83 ha/year (0.32%/year), 50.23 ha/year (0.38%/year), 8.68ha/year (0.07%/year) and 67.31 ha/year (0.51%/year), respectively over an average period of 13 consecutive years (i.e.1995 and 2008). Natural forest and woodland decreased consistently at a rate of -156.46 ha/year (-1.19%/year) and -45.82ha/year (-0.35%/year) over an average period of 13 years (i.e.1995 and 2008) as described in Table 9.

[...]