This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

Ecological Indicators 11 (2011) 1437–1446

Contents lists available at ScienceDirect

Ecological Indicators

journa l homepage: www.e lsev ier .com/ locate /eco l ind

Forest transition in an ecologically important region: Patterns and causes forlandscape dynamics in the Niger Delta

Alex O. Onojeghuo ∗, G. Alan BlackburnLancaster Environment Centre, Lancaster University, Lancaster LA1 4YQ, UK

a r t i c l e i n f o

Article history:Received 3 April 2010Received in revised form13 December 2010Accepted 19 March 2011

Keywords:Niger DeltaDeforestationAfforestationClassificationForest transitionFragmentationMarkov

a b s t r a c t

The influence of human related actives such as oil and gas exploration, intensified logging of trees andover exploitation of forest resources for food, have negatively impacted the once flourishing and ecolog-ically diverse forest system of the Niger Delta region in Nigeria. Relevant information on the transitionalchanges of forested landscapes in the delta is poor compared to other tropical forests such as the Brazilianand Columbian Amazonian forest where numerous research studies have been conducted. Consequently,this study aimed at investigating the spatial extent and rates of forest transition in the Niger Delta regiontaking into consideration the patterns, causes and implications of the landscape dynamics. The studydetermined the spatial extent and rates of forest transition in the study area using remotely sensed datafrom 1986 and 2007. The results indicated that the spatial extent of deforestation, unchanged forest coverand afforestation were 1.38, 2.39, and 1.15 million hectares, respectively, while the annual deforestationand afforestation rates were 0.95 and 0.75% which are high compared to other areas in the humid tropics.The annual rate of change in forest cover was determined as −0.13% indicating an overall reduction inthe spatial extent of forest cover for the entire delta. Changes in the spatial structure of forests wereinvestigated using landscape metrics and the results showed there was a substantial increase in forestfragmentation. The variations in population dynamics and poverty indicators between different statesof the Niger Delta were unable to explain the observed patterns of forest change. Instead, the authorsobserved that the main determinants of forest dynamics were the variations in state forest manage-ment policies and the influence of the oil and gas industry on the economies of the states. High rates ofafforestation were found in states that have limited oil resources and were more economically dependenton forest products, while states with high deforestation rates were found in the main oil-producing partsof the study site. Using the present trend of forest transition dynamics, a 20-year forward simulation wasgenerated using the Markov algorithm. The results concerning forest transition in the study area pointto the urgent need for appropriate environmental policy development and implementation for the NigerDelta region.

© 2011 Elsevier Ltd. All rights reserved.

1. Introduction

The fragile mosaic of aquatic, wetland and terrestrial ecosystemsof the Niger Delta region in Nigeria is faced with a diverse rangeof environmental problems resulting from factors such as agricul-tural land degradation with consequent soil erosion and flooding;shoreline development; changes in local hydrology resulting fromsea level rise; pollution from the oil industry, pipeline constructionand seismic activities; over exploitation of forest resources for foodand construction, illegal felling of trees, agricultural activities andrapid urbanisation (Moffat and Linden, 1995; NDES, 1997; Wanget al., 2005). In particular, deforestation is a major threat affect-

∗ Corresponding author. Tel.: +44 077 6868 6613; fax: +44 015 2451 0269.E-mail address: a.onojeghuo@lancaster.ac.uk (A.O. Onojeghuo).

ing the natural ecological status and biodiversity of the region, aswell as the human population that depends upon resources andservices provided by the ecosystems of the delta for their liveli-hood. The economically viable resources derivable from the NigerDelta forest landscape can be classified into timber and non-timberforest resources. Some of the major timber resources derivablefrom the forests of the Niger Delta are saw-logs, chewing sticks,transmission poles, building and furniture wood/poles, and fuelwood which are obtained from the mangroves, rain forest areas,coastal ridges and fresh water swamps, all of which are being overexploited (NDES, 1997). The non-timber resources are derivablefrom oil palms, raffia palm, bush mangoes (“ogono”), food spices,medicinal roots and leaves, giant snails and many others. The pro-duction of oil from the raffia palms is a huge source of revenue insome states such as Delta and Rivers which are the highest pro-ducers (NDES, 1997). It has been recognized for some time that

1470-160X/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.ecolind.2011.03.017

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

1438 A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446

deforestation is a large-scale problem in the freshwater swampsand barrier island forests of the Niger Delta region with an esti-mated 5–10% of primary mangrove forests having been lost (Moffatand Linden, 1995). Much of the deforestation has been attributedto urban growth and intensified human activities such as loggingand the expansion of agriculture (Moffat and Linden, 1995; NDES,1997). In the pre-colonial and colonial periods the great demand oftropical hardwoods by Europe heralded the threat of deforestationin the Niger Delta region of Nigeria. This led to the introduction ofintensive lumbering activities and tree crop exportation from thelowland forests of the Niger Delta to Europe (Areola, 2001). Therapid increase in the over exploitation of these forest resourcesincreased as timber export and plywood companies came into theheart of the then Bendel state (now known as Delta and Edo states)of the Niger Delta. This was done in order to facilitate the harvestingof timber from forests in Benin, Edo state and Sapele, Delta state.Development activities like road construction, urban expansion andestablishment of industries resulted in the once fertile and flourish-ing lowland forest becoming an over-exploited region with most ofthe forest resources lost to intensified human activity (Moffat andLinden, 1995).

At present, information concerning the current rates of defor-estation for the region is lacking, as, indeed, is knowledge of therates and patterns of afforestation. The Federal Department ofForestry (FDF), which has the responsibility of coordinating forestryrelated issues in Nigeria, has embarked on a number of nationwidestudies in the past through the Forestry Management, Evaluationand Coordinating Unit. Over the last three decades projects havebeen executed to determine the state of the country’s forests, pro-ducing information on land-use and vegetation cover from a varietyof sources. Such studies have demonstrated that deforestation hasbeen particularly prevalent in the types of forests such as man-groves, swamps and riparian forests that dominate the Niger Delta(Areola, 2001; Olaleye and Ameh, 1999; Treveet, 1979). However,Areola (2001) observed that the major problems facing the FDFwere the absence of up-to-date information on the state of forestsin the country and lack of a properly coordinated system to managesuch information, hence resulting in information being distributedacross different organisations in the country. Similarly, Olaleye andAmeh (1999) identified limited coverage of forest inventories andthe absence of up-to-date information across states in Nigeria asmajor problems facing the government in determining the coun-try’s current status of forest changes.

The Niger Delta region is faced with the difficulty of inaccessi-bility during ground surveys thus making most parts of the regionunreachable due to the dense forest cover and difficult undulat-ing terrain. Moreover, the volatile communities present in manyparts of the Niger Delta can compromise the personal safety ofnon-indigenes and prevent in situ surveys. In such circumstances,the use of remotely sensed data offers numerous advantages forproviding forest related information across the study area. Fur-thermore, the possibility of repeated image acquisitions over longperiods of time and with a synoptic view covering large spatialareas, provides a means of monitoring forest dynamics across theentire Niger Delta. Indeed, this has been demonstrated to someextent in a study conducted by Godstime et al. (2007) where Land-sat satellite imagery covering the Niger Delta was used to map thespatial extent of mangrove loss in the region. The study also iden-tified as the primary cause of deforestation urbanisation, dredgingactivities, oil exploration and the spread of the invasive Nypa Palm.

Deforestation can lead to intensified fragmentation resultingin the reduction of forest patches and increased isolation. Stud-ies aimed at quantifying the nature and consequences of forestfragmentation have been conducted and vital results obtained. Forexample, it has been found that the tropical forests of the Amazonare subject to accelerating deforestation rates resulting in forest

fragmentation and changing patterns of ecosystem loss (Lauranceet al., 2000; Lima and Gascon, 1999). The study of Armenteraset al. (2006) on the Colombian Amazon examined the pattern ofecosystem diversity, deforestation and fragmentation through thetemporal and spatial analysis of biotic and abiotic data using remotesensing and geographic information system. The pattern of defor-estation and annual rates of deforestation for different parts ofthe region were investigated. The annual deforestation rates were3.73 and 0.97% in two areas with high human population densitieswhile for relatively unpopulated areas the rates were 0.31, 0.23, and0.01%. At present similar information on the level of forest transi-tion and forest landscape fragmentation is lacking for the NigerDelta.

The aim of this study was to investigate rates of forest transitionin the Niger Delta region taking into consideration the patterns,causes and implications of landscape dynamics in the delta. Theobjectives were as follows:

i. to determine the spatial extent and rates of deforestation,afforestation and forest change in the Niger Delta using remotelysensed imagery,

ii. to investigate the effects of forest transition on the spatial struc-ture of forest landscapes in the Niger Delta,

iii. to examine the extent to which socio-economic factors influ-ence the processes of forest transition, and

iv. to simulate likely future changes in forest cover of the NigerDelta.

2. Materials and methods

2.1. Study area

The study area was the Niger Delta (Fig. 1). Geographicallythe Niger Delta is taken to have its apex at Aboh 5◦33′49′′N,6◦31′38′′E which bifurcates into two main distributaries, the Nunand Forcados. The southernmost tip of the Delta is at Palm Point(4◦16′22′′N, 6◦05′27′′E) south of Akassa and at the estuary of NunRiver. The Niger Delta region extends from the Benin river estu-ary (5◦44′11′′N, 5◦03′49′′E) in the west to the Imo river estuary(4◦27′16′′N, 7◦35′27′′E) in the east and has a total land area ofapproximately 25,900 km2 or approximately 2.8% of Nigeria’s totalland area (NDES, 1997). The natural vegetation of the Niger Deltais classified into five types which are as follows: moist lowlandrainforests covering the upper riverine floodplains and Sombreiro-Warri plains; freshwater swamp forest covering the lower riverinefloodplains and along the river valleys; the mangrove zone com-prising the mangrove forests and mangrove swamps located in theupper tidal and lower tidal zones, respectively; saltmarsh and tidalmudflats along the shorelines; and coastal forests and thickets onthe barrier sand ridges (NDES, 1997). In the Niger Delta populationdistribution and settlement pattern are determined by the avail-ability of dry land within the mangrove swamp dominated zone.Most of the Niger Delta is characterized by massive and continuousswamps dotted by islands of dry land. Due to the narrow fringe-like islands dominant in the region land reclamation (through sandfilling) is a common practice in order to accommodate the grow-ing population (Moffat and Linden, 1995). Consequently, the largestsettlements of the Niger Delta are locations further away from themangroves such as Benin, Sapele, Warri, Port Harcourt and Ughelli(NDES, 1997).

2.2. Mapping deforestation and afforestation

For this study the analysis of remotely sensed data wasundertaken in four stages namely: data acquisition and

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446 1439

Fig. 1. Map of the Niger delta region with the inserts of the map of Africa and Nigeria showing the Niger delta.

processing, image classification, accuracy assessment andchange detection. These are explained in the followingsections.

2.2.1. Data acquisition and image pre-processingThe Landsat Thematic Mapper (TM) satellite imagery used for

the study was obtained from the Global Land Cover Facility (GLCF)website (http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp)which is managed by the University of Maryland and fundedby the National Aeronautics and Space Agency (NASA) in theUnited States. The Landsat TM data was acquired on the 19th ofDecember 1986. A NigeriaSat1 image covering the same studyarea was acquired on the 20th of January 2007 by the DisasterMonitoring Constellation International Imaging (DMCii). Theimages were acquired during the dry season in order to limit cloudcover.

The rectification and restoration of the satellite image cover-ing the study area was vital to the accuracy of the final productas the imagery in its raw state contained geometric distortions.The images were geometrically corrected using a polynomial geo-metric model in ERDAS imagine (ERDAS Inc., 2008). To effectivelymaximise the information contained in the bands of the satelliteimagery atmospheric correction was applied to remove the effectsof atmospheric particles caused by absorption and scattering ofradiation from the earth surface during acquisition (Lu et al., 2002).The presence of haze noise was observed and removed using thehaze reduction command in ERDAS Imagine (ERDAS Inc., 2008). Thepre-processing and further analysis was performed using ERDASImagine (version 9.2). The spectral radiance of each band containedin the imagery was calculated and converted to reflectance whichwas used as an input for image classification. All the bands of theimagery used in the study were resampled to a pixel value of 32.0 inorder to normalize the spatial scale differences between the bandsof both datasets.

The conversion of raw digital numbers to at-satellite reflectancevalues was done according to methods outlined in the Landsat 7 Sci-ence Data Users Handbook (Irish, 2000), similar to related projects(Huang et al., 2002). The raw digital numbers of the Landsat TMand NigeriaSat1 images were converted to at-satellite values toremove all forms of noise that were introduced due to atmosphericeffects, change views and illumination geometry as well as instru-ment errors (Huang et al., 2002; Iqbal, 1983). This normalised theimpact of illumination geometry. Eqs. (1)–(3) below were used inthe computation.

L(i) = DN(i) × gain(i) + bias(i) (1)

L(u) =(

DN(u)

gain(u)

)+ bias(u) (2)

R(i) = 3.1415926 × L(i) × d2

Esun × sin �(3)

where i is the Landsat TM band number; u is the NigeriaSat1band number; L(i)/(u) are at-satellite radiance of Landsat and Nige-riaSat1 images; gain(i)/(u) and bias(i)/(u) are band specific valuesobtained from the metadata or HDR files of both datasets; d isthe normalised Sun-earth distance calculated from the eccentricitycorrection factor E0 (d2 = 1/E0) (Iqbal, 1983), Esun is the mean solarexoatmospheric irradiance (Irish, 2000), and � is the sun elevationas provided in the image metadata file.

2.2.2. Image classificationThe Iterative Self Organising Data Analysis (ISODATA) algorithm

in ERDAS Imagine software (version 9.2) was used to perform anunsupervised classification of both images. The ISODATA cluster-ing method uses the minimum spectral distance formula to formclusters. It begins with either arbitrary cluster means or means ofan existing signature set (ERDAS Inc., 2008). Each time the clus-tering repeats, the means of these clusters are shifted. The new

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

1440 A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446

cluster means are used for the next iteration. The ISODATA utilityrepeats the clustering of the image until either a maximum num-ber of iterations have been performed, or a maximum percentageof unchanged pixel assignments reached between two iterations.For this study the unsupervised ISODATA classification techniquehad 30 classes, confidence threshold of 0.99, and 200 maximumiterations. The 30 classes were subsequently categorised into fivemajor landcover classes namely: water, forest, bare soil, settlementand sand deposits.

2.2.3. Accuracy assessmentDue to the difficulty of accessing certain swamp terrain of the

Niger Delta, a preselected site named Onya situated in Delta statewhich had been visited by the researcher was selected. The dataused for training and testing image classifications were collectedusing a combination of methods and sources namely: DGPS fieldsurveys covering Onya, a digitised 1: 50,000 topographic sheet cov-ering Onya dated April 1984 and a high resolution orthorectified 1meter IKONOS image dated January 2006 (which was closest to theNigeriaSat1 image used). The accuracy assessment was applied tothe classified maps showing the following land-uses: water, for-est, bare soil, settlement and sand deposits. This was accomplishedusing Accuracy Assessment command in ERDAS Imagine software.The results of the accuracy assessment for NigeriaSat1 image was90% for the overall classification accuracy and 0.87 for the Kappastatistic.

2.2.4. Change detectionIn order to determine the spatial extent of changes in the

forested areas the classified images were recoded to define theforested areas as one class and all other land covers types werecombined into a non-forest class. A post-classification comparisonwas adopted to determine the changes using the Change detec-tion tool in ERDAS Imagine software. The forest transition mapsshowing deforested, afforested and unchanged forest landscapeswere based on the difference between the forested area cover inthe baseline (1986) and assessment (2007) maps for the studyarea.

2.3. Computing average annual rates of forest cover transition

The average annual rates of deforestation, afforestation andchange in total forest cover were evaluated using spatialdetails of the forest cover area in 1986 and 2007. This wasbased on the assumption that the rates of annual deforesta-tion, afforestation and change in total forest cover will notbe constant (Armenteras et al., 2006; Lele and Joshi, 2008).Annual forest transitions were computed using the equations

below:

Annual deforestation rate = (log Ft2 − log(Ft1 − B)) × 100t2 − t1

(4)

Annual afforestation rate = (log Ft2 − log(Ft1 − C)) × 100t2 − t1

(5)

Annual rate of total forest cover change = (log Ft2 − log Ft1) × 100t2 − t1

(6)

where F is the forested area in hectares, at t1 (1986) and t2 (2007);B is the area deforested between 1986 and 2007; and C is the areaafforested between 1986 and 2007.

2.4. Forest cover change analysis and predicted dynamics

The extent of forest fragmentation was determined usingFRAGSTATS software, a computer software program designed tocompute a wide variety of landscape metrics for categorical mappatterns (McGarigal et al., 2002). The metrics used in the studyincluded the total area, percentage of landscape, number of patches(NP), mean patch area (MPA) and patch density (PD). The PD isequivalent to the number of patches of a particular patch type (i.e.forest patches) divided by the total landscape area.

Markov chain analysis is a technique for predictive change mod-eling and is based on past occurrences. In this study the Markovchain and probabilities matrix was used to evaluate the futurestatus of forest cover in the Niger Delta outputting as results thepercentage and probability of forest land use converting to non-forest and vice versa in the future. The Markov module in IDRISIAndes 15 (Eastman, 2006) was used for this procedure to perform a20-year forward simulation. The example cited by Eastman (2006)used two different time periods (1971 and 1985) to project lan-duse change into the future (1999) using the MARKOV module inIDRISI Andes 15. The module analyses a pair of land cover imagesand outputs a transition probability matrix and a set of conditionalprobability images (Eastman, 2006). The transition probabilitymatrix records the probability of each land cover type changing toanother landcover type, while the transition areas matrix recordsthe number of pixels that are expected to change from each landcover type to each other land cover type over the specified numberof time units. In the present study the conditional probability val-ues showed the probability of forest land-cover being converted tonon-forest and vice versa. The Markov chain analysis was executedon a state by state basis and the results outputted into ArcGIS forfurther analysis.

Table 1Landscape measures from 1986 to 2002/03 of eight states in the Niger Delta region of Nigeria.

State Forest landscapearea (ha)

Percentage of forestlandscape (%)

Number of patches Patch Density (no.per 100 ha)

Mean patch area(ha)

1986 2007 1986 2007 1986 2007 1986 2007 1986 2007

Abia 192,968 287,092 39.73 59.11 46,741 40,604 9.62 8.36 4.13 7.07Akwa Ibom 383,829 533,071 58.54 81.30 36,525 18,524 5.57 2.83 10.51 28.78Bayelsa 791,441 659,829 84.54 70.48 9,150 42,662 0.98 4.56 86.50 15.47Cross River 467,276 460,433 70.67 69.63 19,426 37,055 2.94 5.60 24.05 12.43Delta 834,676 516,956 56.08 34.74 106,942 140,072 7.19 9.41 7.80 3.69Edo 176,804 31,104 34.96 6.15 71,228 25,794 14.09 5.10 2.48 1.21Imo 267,720 364,085 52.14 70.90 42,986 28,190 8.37 5.49 6.23 12.92Rivers 650,422 681,311 62.77 65.75 40,088 51,339 3.87 4.95 16.22 13.27

Niger delta 3,765,136 3,533,880 59.93 56.25 373,086 384,240 5.93 5.63 10.09 9.20

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446 1441

3. Results and discussion

3.1. Changes in forest cover

The total forest cover of the Niger Delta after image classifica-tion was 3,765,136 ha in 1986 and 3,533,880 ha in 2007 (Table 1).The total area of the forest and non-forest areas was 6,282,574 ha.The percentage of total forest cover in the Niger Delta declinedfrom 60% in 1986 to 56% in 2007. In terms of percentages, thestates with the highest forest cover were Bayelsa in 1986 (85%)and Akwa Ibom in 2007 (81%) (Table 1). The lowest forest coverwas recorded for Abia in 1986 (40%) and Edo in 2007 (6%). Between1986 and 2007 the percentage of forest cover increased for Abia,Akwa Ibom, Imo and Rivers states and declined for the Bayelsa,Cross River, Delta, and Edo states. The results showed that all thestates in the Niger Delta had some form of variation in forest lossand gain over the 21 years time interval. Table 2 shows the resultsof forest cover change (indicating the areas that were deforested,unchanged and afforested) and the annual rates of change in theforest cover. The greatest loss in forest cover was in Delta statewith a massive loss of 469,731 ha, 8% of the entire study area. Theannual rates of deforestation varied between 4.94 and 0.45% acrossstates which are higher than the rates observed in the ColumbianAmazon which ranged between 3.73 and 0.01% (Armenteras et al.,2006). The total forest cover lost to deforestation in the entire NigerDelta was 1,381,217 ha, representing a large transition across 22%of the total landscape area. The annual deforestation rate for theNiger Delta from 1986 to 2007 was 0.95% while the afforestationrate was 0.75%. These rates exceed those reported for humid tropi-cal forests globally and in Latin America, Africa and Southeast Asia(Achard et al., 2002). The annual rate of change in forest cover forthe entire Niger Delta from 1986 to 2007 was −0.13% indicative of ageneral trend in forest cover decline. Bayelsa, Cross River, Edo andDelta states experienced forest cover decline, with Edo and Deltabeing the highest, while Abia, Akwa Ibom, Imo and Rivers statesexperienced a forest cover increase. Table 1 shows the deforestedand afforested landscape area for states in the Niger Delta from1986 to 2007.

3.2. Changes in forest spatial structure

Changes in the forest landscape structure from 1986 to 2007were observed in all eight states of the Niger Delta. The changes inNP, PD and MPA which are measures of the extent of subdivisionor fragmentation of the forests are shown in Table 1. The total NPincreased in Bayelsa, Cross River, Delta and Rivers states, while itdecreased in Abia, Akwa Ibom, Edo and Imo states (Table 1). PDshowed an opposite change to the NP results for each of the states.Overall the NP for forest cover in the Niger Delta for 1986 and 2007were 373,086 and 384,240, respectively, which may not appear tobe a large difference overall, but the changes in NP tend to be con-centrated in particular states (Table 1). For Bayelsa, Cross River,Delta, Edo and Rivers states there was a decline in the MPA from1986 to 2007, a further indication that fragmentation had occurredin the five states. However, Abia, Akwa Ibom, and Imo states had anincrease in MPA indicating an overall expansion of existing forestpatches. The MPA for the entire Niger Delta decreased from 10.1 hato 9.2 ha between 1986 and 2007. Overall, these spatial metricsdemonstrate that for most states in the Niger Delta, forest coverhas declined and the remaining patches have become more frag-mented. However, there is a group of states where forest coverhas increased and this is mostly through the enlargement of for-est stands rather than the creation of new patches. Such variationsbetween states in the dynamics of forest spatial structure are likelyto generate different environmental effects. Nevertheless, the dom-inant process of forest cover decline and fragmentation across the Ta

ble

2R

esu

ltof

chan

ged

etec

tion

show

ing

spat

ialc

han

ges

ofva

riou

sfo

rest

cove

rty

pes

inth

eN

iger

Del

taov

er21

year

sin

terv

al.

Stat

eD

efor

este

d(h

a)U

nch

ange

dfo

rest

(ha)

Aff

ores

ted

(ha)

Def

ores

ted

(%)

Un

chan

ged

(%)

Aff

ores

ted

(%)

An

nu

ald

efor

esta

tion

rate

(%)

An

nu

alaf

fore

stat

ion

rate

(%)

An

nu

alra

teof

chan

gein

fore

stco

ver

(%)

Abi

a10

0,16

594

,803

192,

289

25.8

724

.48

49.6

51.

5111

.68

0.82

Akw

aIb

om75

,190

308,

639

224,

431

12.3

650

.74

36.9

00.

451.

820.

68B

ayel

sa17

8,72

261

2,71

947

,110

21.3

173

.07

5.62

0.53

0.13

−0.3

8C

ross

Riv

er13

2,15

933

5,11

712

5,31

522

.30

56.5

521

.15

0.69

0.65

−0.0

3D

elta

469,

731

364,

945

152,

011

47.6

136

.99

15.4

11.

710.

42−0

.99

Edo

160,

544

16,2

6014

,844

83.7

78.

487.

754.

940.

18−3

.59

Imo

81,9

0218

5,81

917

8,26

718

.36

41.6

639

.97

0.76

2.27

0.64

Riv

ers

182,

805

467,

616

213,

694

21.1

654

.12

24.7

30.

680.

820.

10

Nig

erD

elta

1,38

1,21

72,

385,

919

1,14

7,96

121

.98

37.9

818

.27

0.95

0.75

−0.1

3

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

1442 A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446

Table 3Socio-economic indicators of eight states in the Niger Delta region of Nigeria.

State Poverty headcount (%) Population Population density(inhabitants per sq. km)

Population growthrate (%)

1963 1991 2006 1963 1991 2006

Abia 82.46 * 2,338,487 2,833,999 * 477 578 2.7Akwa Ibom 65.36 * 2,409,613 3,920,208 * 349 568 3.4Bayelsa 95.57 * 1,067,793 1,703,358 * 118 188 2.9Cross River 81.16 3,478,000 1,911,297 2,888,966 113 88 133 2.9Delta 84.25 * 2,590,491 4,098,391 * 151 240 3.2Edo 83.34 2,461,000 2,172,005 3,218,332 68 113 168 2.7Imo 77.78 3,673,000 2,485,635 3,934,899 361 470 744 3.2Rivers 72.60 1,719,000 4,309,557 5,185,400 88 408 490 3.4

Niger delta * 11,331,000 19,284,878 27,783,553 *

* signifies the term N/A (not applicable).

delta as a whole is likely to have deleterious effects on the provi-sion of a wide range of ecosystem services. While fragmentationcan lead to greater habitat diversity, studies in other tropical forestenvironments, have clearly demonstrated that fragmentation leadsto adverse effects on range-restricted endemic species (Martin andBlackburn, 2010), overall species diversity (Fahrig, 2003), carbonsequestration (Groeneveld et al., 2009) and controls on the hydro-logical cycle (Giambelluca, 2002).

3.3. Socioeconomic drivers and processes of forest landscapetransitions

Part of this study aimed at investigating if there were rela-tionships between specific socio-economic indicators and thecharacteristics of forest transition in the Niger Delta region. Inthis regard data on the poverty ratings percentage and populationstatistics for all eight states in the Niger Delta were obtained fromthe office of the National Bureau of Statistics in Nigeria. The conceptof measuring poverty level has a number of factors contributing tothe proper definition of this term. Poverty is not the lack of incomealone but the absence of basic requirements such as health facil-ities, education and employment (NBS, 2005, 2006; NDES, 1997).Recently powerlessness, isolation, vulnerability and social exclu-sion have been included in having a proper definition of poverty.The poverty headcount percentages recorded by the Office of statis-tics (2004) indicate high levels of poverty across the Niger Deltabut with considerable variations between states (Table 3). Havingregressed the annual afforestation and deforestation rates againstthe poverty headcounts between states it was revealed that therewas no direct relationship between poverty and the rates of defor-estation or afforestation across the states, indicating that there areother significant drivers of forest transition in the region.

The population figures from the 1963, 1991 and 2006 cen-suses conducted by the National Population Commission indicatea steady increase in the population of states comprising the NigerDelta (Table 3). In 1963 Rivers and Bayelsa states (formerly knownas Rivers state) had a population of about 1.7 million peoplecompared to 2.5 million, 3.5 million and 3.7 million inhabitantsof Delta and Edo (formerly Bendel), Cross River and Akwa Ibom(formerly Cross River) and Imo and Abia (formerly Imo) states,respectively (NBS, 2006). Results of the 1996 and 2006 censusesrevealed a rapid increase in population for each state. The high-est was recorded in Rivers state having an exponential growthrate of 3.4%, followed by Delta and Imo states each having 3.2%.Akwa Ibom, Bayelsa and Cross River each had growth rate val-ues of 2.9% while Abia and Edo had the smallest growth rate of2.7%. Again, our analysis showed that there was no direct statisti-cal correlation between population dynamics and the variation inforest transitions between states. However, it is apparent that eco-

nomically driven variations in population and forest practices areimportant.

Prior to the 1970s migration into the Niger Delta was deter-mined by the agricultural, fishing and trading needs of inhabitantsin and around the region (NDES, 1997). However, after the dis-covery of oil in the Niger Delta the rate of migration increased aspeople from all over the country came to the urban areas of the corestates of the region (particularly Rivers and Delta states) seekingjobs in oil companies. This pattern of migration was responsiblefor the massive growth rate witnessed in Rivers state (Table 3).Most of the major oil companies in Nigeria have their headquar-ters in Port Harcourt, capital of Rivers state. The inhabitants ofthe state rely more on the revenue derivable from the oil indus-try and trading activities. Port Harcourt has a history of beinga major merchant port and is the center of a variety of indus-try with exports such commodities as petroleum, coal, tin, palmproduce, cocoa and groundnuts. Given the limited availability ofland in the mangrove dominated region of the core states, thepopulation density on the habitable areas has increased greatlyresulting in intensified pressure and over exploitation of the for-est resources in the region (WAD, 1990; Moffat and Linden, 1995).It should be noted that most of the states with high afforestationrates are further away from the core states of the Niger Delta, andlack the financial resources of the oil rich states. Fig. 2 shows themaps of forest cover derived from remotely sensed data in 1986and 2007, together with map of forest transitions that have takenplace in the Niger Delta over the 21 years period. At a glance onecan see the high transition in forest cover in Delta and Edo stateswhich were mostly affected by the trend of forest loss, while statessuch as Imo, Abia and Akwa Ibom have experienced considerableafforestation.

Using contemporary aerial photographs to provide greater spa-tial detail, Fig. 3 provides examples of forest transition hotspots,showing deforested and afforested sites. The influence of humanactivities such as construction activities and intensified loggingwithout replacement of felled trees has lead to the fragmenta-tion of forest landscapes in the region (Fig. 3a–c). The patternof deforestation in the Niger Delta is influenced by access routessuch as roads, tracks and water ways; settlement expansion andconstruction activities such as power base stations; intensifiedexploration of crude oil by oil companies and seismic companies;and exploitation of forest resources for building, furniture makingand export. A similar situation was observed in the tropical forestsof the Colombian Amazon where the patterns of deforestationwere influenced by the river transportation network, demo-graphic pressure, quality of life and population density (Armenteraset al., 2006). In the Niger Delta some states have embarked onconserving forests and growing forest and oil palm plantations(Fig. 3d–f).

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446 1443

Fig. 2. Classified forest cover and forest transition hotspot map of the Niger delta from 1986 to 2007 (a: classified forest map for 1986, b: classified forest map for 2007, c:afforested, unchanged and deforestation areas hotspot map of the study area).

Table 4Results of the Markov chain analysis: transition probability and transition area matrix for the Niger Delta.

State Land cover Transition probabilities Transition areas (ha)

Forest Non-forest Forest Non-forest

AbiaForest 0.47 0.53 135,969 151,123Non-forest 0.67 0.33 133,680 64,949

Akwa IbomForest 0.80 0.20 428,482 104,588Non-forest 0.83 0.17 101,398 21,251

BayelsaForest 0.77 0.23 510,828 149,001Non-forest 0.33 0.67 89,935 186,457

Cross RiverForest 0.72 0.28 330,865 129,568Non-forest 0.64 0.36 129,084 71,709

DeltaForest 0.44 0.56 226,029 290,927Non-forest 0.23 0.77 225,909 745,371

EdoForest 0.09 0.91 2,861 28,243Non-forest 0.05 0.95 21,420 453,152

ImoForest 0.69 0.31 252,448 111,637Non-forest 0.73 0.27 108,623 40,802

RiversForest 0.72 0.28 491,988 189,323Non-forest 0.55 0.45 194,365 160,587

Niger Delta total transition of forest landscape: 2,379,470 1,154,410Niger Delta total transition of non-forested landscape: 1,004,414 1,744,278

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

1444 A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446

Fig. 3. Aerial photograph (2007) showing deforestation and afforestation hotspots in the Niger delta (a: the power station has lead to the fragmentation of the forest landscapedistorting its initial ecosystem stability; b: the influence of access routes (roads and tracks) to forest fragmentation in the NW part of Okpai Delta state; c: the influenceof settlement expansion and road access on deforestation; d: afforested landscape in Abia state; e: afforested landscape in Ovia community Delta state; and e: forestedlandscape situated in Umuahia Abia state).

3.4. Forward simulation of forest cover

A forward simulation was performed order to model the futurestatus of the forests in the Niger Delta taking into consider-ation the pattern of forest transition recorded between 1986and 2007. Table 4 shows the transition probabilities for each ofthe states together with the predicted areas of each state thatwill remain forested or non-forested or undergo conversion toforest or non-forest. Table 4 demonstrates graphically the vari-ations between states in (a) the transition probabilities and (b)the total area of deforestation and afforestation. Fig. 4 showsthe transition probability maps of forest cover in 2027 derived

from remotely sensed data in 1986 and 2007. The results indi-cate that the highest probability of deforestation is in Edo statebut because of the small area of remaining forest in 2007, thespatial extent of deforestation predicted for 2027 is small. Deltastate is predicted to experience the largest spatial extents of bothdeforestation and afforestation, while most other states will haveapproximately the same areas that undergo deforestation andafforestation. These changes will result in large rates of transi-tion of the forested landscape across the Niger Delta as a whole.As shown in Table 4, the total area of deforestation is predictedto 1,154,410 ha which amounts to 18% of the total geographicalarea of the delta. The total area to become afforested landscape

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446 1445

Fig. 4. Transition probability maps of states in the Niger delta showing forward simulation of forested areas in 2027.

is 1,004,414 ha which is 16% of area of the delta. Hence, underpresent forest management regimes, some 34% of the area of thedelta is predicted to undergo some form of forest transition by2027.

4. Conclusion

This study has identified the spatial extent of forest transitionin the Niger Delta region of Nigeria using remotely sensed data.The annual rate of deforestation, afforestation and forest coverchange for the entire Niger Delta region was determined as 0.95,0.75 and −0.13 percent, respectively. These represent rates of for-est transition which are high compared with those found in manyprevious studies in tropical forest environments. There were con-siderable variations between the different states of the delta regionin the characteristics of forest transition. Deforestation hotspotswere found in the states of Bayelsa, Cross River, Delta and Edo whileAbia, Akwa Ibom, Imo and Rivers states experienced high rates ofafforestation. Landscape spatial metrics indicated that the forestsof the Niger Delta have undergone high levels of fragmentation,which, as observed in other tropical forest regions, is likely to resultin the degradation of ecosystem services provided by the forestedlandscape.

The results indicated that there were no direct relationshipsbetween the characteristics of forest transition and poverty head-counts or population growth across the states of the Niger Delta.However, forest transitions can be accounted for by the differentmeans of managing forest resources implemented in the states inresponse to the variations in the balance between their economicdependence on oil or forest-based resources. Our simulations of

future changes indicated that, based on current patterns of land-scape management, a substantial proportion of the Niger Delta willundergo some form of forest transition over the coming decades.Given the likely ecological and environmental impacts of this large-scale trend, urgent attention needs to be given to the developmentand implementation of more sustainable management policies andpractices.

Acknowledgements

This study was made possible by the sponsorship of thePetroleum Technology Development Fund (PTDF) provided by theGovernment of Nigeria. Special thanks to the DMCii for providingthe NigeriaSat1 imagery used for the study.

References

Achard, F., Eva, H.D., Stibig, H.J., Mayaux, P., Gallego, J., Richards, T., Malingreau, J.P.,2002. Determination of deforestation rates of the world’s humid tropical forests.Science 297, 999–1001.

Areola, O., 2001. Deforestation in Nigeria. In: Dhirendra, V.K. (Ed.), Deforestation,Environment, and Sustainable Development: A Comparative Analysis. Westport,CT/London, Praeger/Greenwood.

Armenteras, D., Rudas, G., Rodriguez, N., Sua, S., Romero, M., 2006. Pattern and causeof deforestation in Columbian Amazon. Ecological Indicators 6, 353–368.

Eastman, J.R., 2006. IDRISI Andes: Guide to GIS and Image Processing (Version 15).Clark Labs, Clark University, Worcester/USA.

ERDAS Inc., 2008. ERDAS Imagine: ERDAS Field Guide (Volume 1). ERDAS Inc.,Atlanta.

Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annual Review ofEcology, Evolution and Systematics 34, 487–515.

Giambelluca, T.W., 2002. Hydrology of altered tropical forest. Hydrological Processes16 (8), 1665–1669.

Author's personal copyJournal Identification = ECOIND Article Identification = 812 Date: May 6, 2011 Time: 7:31 pm

1446 A.O. Onojeghuo, G.A. Blackburn / Ecological Indicators 11 (2011) 1437–1446

Godstime, J., Adegoke, J., Saba, E., Nwilo, P., Akinyede, J., 2007. Satellite-Based Assess-ment of the Extent and Changes in the Mangrove Ecosystem of the Niger Delta.Marine Geodesy 30, 249–267.

Groeneveld, J., Alves, L.F., Bernacci, L.C., Catharino, E.L.M., Knogge, C., Metzger, J.P.,Pütz, S., Huth, A., 2009. The impact of fragmentation and density regulationon forest succession in the Atlantic rain forest. Ecological Modelling 220 (19),2450–2459.

Huang, C., Wylie, B., Yang, L., Homer, C., Zylstra, G., 2002. Derivation of a tasselled captransformation based on Landsat 7 at-satellite reflectance. International Journalof Remote Sensing 23, 1741–1748.

Iqbal, M., 1983. An Introduction to Solar Radiation. Academic Press.Irish, R.R., 2000. Landsat 7 Science Data User’s Handbook. National Aeronautics and

Space Administration.Laurance, W.F., Delamonica, P., Laurance, S.G., Vasconcelos, H.L., Lovejoy, T.E.,

2000. Conservation: rainforest fragmentation kills big trees. Nature 404,836–1836.

Lele, N., Joshi, P.K., 2008. Analyzing deforestation rates, spatial forest coverchanges and identifying critical areas of forest cover changes in North–EastIndia during 1972–1999. Environmental Monitoring and Assessment 156,159–170.

Lima, M.G., Gascon, C., 1999. The conservation value of linear forest remnants incentral Amazonia. Biological Conservation 91, 241–247.

Lu, D., Mausel, P., Brondizio, E., 2002. Assessment of atmospheric correction meth-ods applicable to Amazon basin LBA research. International Journal of RemoteSensing 23, 2651–2671.

Martin, T.E., Blackburn, G.A., 2010. Impacts of tropical forest disturbance upon avi-fauna on a small island with high endemism: implications for conservation.Conservation and Society 8, 127–139.

McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., Weng, Q., 2002. FRAGSTATS: Spa-tial Pattern Analysis Program for Categorical Maps. Available at the followingwebsite http://www.umass.edu/landeco/research/fragstats/fragstats.html.

Moffat, D., Linden, O., 1995. Perception and reality: assessing priorities forsustainable development in the Niger River Delta. Ambio Stockholm 24,527–538.

NBS, 2005. National bureau of statistics report. In: Akinyosoye, V.O. (Ed.), SocialStatistics in Nigeria 2005. National Bureau of Statistics, Nigeria.

NBS, 2006. Statistical Fact Sheet. National Bureau of Statistics.NDES, 1997. Niger Delta Environmental Survey: Final Report Phase. 1. Environ-

mental and Socio-Economic Characteristics. Environmental Resource ManagersLimited.

Olaleye, O.A., Ameh, C.E., 1999. Forest Resource Situation Assessment of Nigeria.Forestry Statistics and Data Collection. EC-FAO Partnership Programme(1998–2002) – Project GCP/INT/679/EC.

Treveet, J.W., 1979. Mapping Nigeria’s forest by radar. West African TechnicalReview, 71–75.

WAD, 1990. Towards the Development of an Environmental Action Plan for Nigeria.West Africa Department, World Bank, Washington, DC.

Wang, C., Qi, J., Cochrane, M., 2005. Assessment of tropical forest degradation withcanopy fractional cover from Landsat ETM+ and IKONOS imagery. Earth Inter-actions 9, 1–18.