Forest cover changes in Cte dIvoire and Upper Guinea

Introduction

Of the forest that covered West Africa at the beginning of the 20th century, the large majority has disappeared and the Taï National Park forest in Côte d'Ivoire is now the last important one.

Notwithstanding this rampant deforestation, it is necessary to remember that the forest was not always present throughout the centuries: forests took over savanna zones, and vice versa (Guillaumet & Adjanohoun 1971, Maley 1996, van Rompaey 1993). More recently, extensive savanna zones got forested during the 1850 to 1920 period (Richards 1996, Fairhead & Leach 1998). Because of the forest's natural cycle of regression and progression, it is difficult to speak of an original or primary forest. This is especially the case for the moist semi-deciduous forest domain (Hawthorne 1996) and for the savanna woodland transitional zone (Spichiger & Blanc-Pamard 1973, Spichiger 1975). However, we need to use these terms to differentiate this state from recently developed secondary forests.

The FAO/UNEP (1981) study on the state of tropical forest in 76 countries showed that on continental or national scales, deforestation was very high between 1975 and 1980. In order to understand the need to conserve certain forests and the impact of deforestation on environmental and economic changes, we must analyse the actual situation of forest cover. Such an analysis is needed for the development of a management policy for the forests at the sub-continental or national level.

In this chapter we analyse in detail the changes in forest cover over the last ten years for Côte d'Ivoire, a country with a long history of forest exploitation (Meniaud 1922, Aubréville 1958, Adjanohoun et al. 1966, Lanly 1969). Since the 1970s, the Ivorian forest has the highest rate of deforestation among tropical countries, 6.5% forest loss per year, compared to a global average of 0.5% per year (FAO/UNEP 1981). For the 1981-1990 period, the average rate of deforestation in Côte d'Ivoire has been assessed at 7.6% per year (850 km2/yr). In 1990 only 11,230 km2 of dense forests were remaining in Côte d'Ivoire, from an original forest cover of 150,000 km2 (FAO 1993, but see Fairhead & Leach 1998), approximately 7.5%.

Areas with a formal protection status have become essential to the conservation of natural settings. Except for

Figure 2.1 Azagny National Park in 2000 consists of forests and swampy savannas. Although this is a national park, the borders are strongly degraded by agricultural activities. A full-colour version of this figure can be found in Appendix 1.

national parks, the purpose of most of these areas is a combination of conservation and production and exploitation of wood, as is the case for the "forêts classées" of Côte d'Ivoire or the forest reserves of Ghana. Although these areas have played a significant role in the preservation of forest, in many cases forest management plans have forgotten to take the rural setting of these forests into consideration. For the sake of improved environmental management, forestry policies must integrate rural and state domains.

A global approach to deforestation, using low resolution images, allows us to acquire information on the whole region and to identify large remaining forests which are the more important for conservation. But in west tropical Africa, the high degree of fragmentation and the complexity of the landscape mosaic is difficult to resolve (Paivinen et al. 1992, Husson 1995). It is necessary to use high resolution images such as LANDSAT in order to describe all the parameters needed to evaluate the extent of deforestation. These parameters are the distribution of small forest patches, the dynamics of secondary forests, the density of cash crop cultivation, and the density of cleared areas bordering on or in protected areas (Chatelain et al. 1996b).

Figure 2.2 Forest cover in West Africa. (A) Cover according to NOAA-AVHRR satellite images (Paivinen et al. 1989). The eastern part of Liberia and western part of Côte d'Ivoire, including Taï National Park, are the last large forested areas that remain. (B) Cover according to the "eco-regions" map of Olson & Dinerstein (1998) representing the forest cover zonation with the limits in 1912 according to Chevalier (1920) and the limits in 1923 following Shantz & Marbut (1923). The variability seems to be due to the personal view of the authors rather than to changes in forest cover between 1912-23. A full-colour version of this figure can be found in Appendix 1.

Figure 2.2 Forest cover in West Africa. (A) Cover according to NOAA-AVHRR satellite images (Paivinen et al. 1989). The eastern part of Liberia and western part of Côte d'Ivoire, including Taï National Park, are the last large forested areas that remain. (B) Cover according to the "eco-regions" map of Olson & Dinerstein (1998) representing the forest cover zonation with the limits in 1912 according to Chevalier (1920) and the limits in 1923 following Shantz & Marbut (1923). The variability seems to be due to the personal view of the authors rather than to changes in forest cover between 1912-23. A full-colour version of this figure can be found in Appendix 1.

Figure 2.3 Forest cover in Côte d'Ivoire in 1955-58 according to the vegetation map (scale 1:500,000) of Guillaumet & Adjanohoun (1969) and limits of the vegetation domains following Monnier (1983). A full-colour version of this figure can be found in Appendix 1.

Figure 2.4 Forest cover in Côte d'Ivoire in 1993 (Dao 1999) based on NOAA-AVHRR images. The Taï forest area represents at least 40% of the total forest area of Côte d'Ivoire. Also the forest area south of Abengourou (the classified forests of Yaya-Bossématié-Mabi) is very important. A full-colour version of this figure can be found in Appendix 1.

In this chapter we will report on forest areas, and their changes, at several scales of resolution. Firstly, we will describe the total forest areas in West Africa. Secondly, we will analyse the total forest cover of Côte d'Ivoire, in particular the total protected area. Thirdly, we will analyse this for the region of Abidjan, as an example for the country. Lastly, we will analyse forest fragmentation in detail for eight blocks of 20 X 20 km in that area.

Method

Study area

We studied the changes in forest cover of the dense forests on four scales.

The first scale is the level of West Africa. This takes into account all the countries of West Africa that had part of their territory covered with rainforest and comprises the phytogeographical zone of Upper Guinea from West Senegal to Nigeria (Figure 2.2). The forested zone probably covered between 56,600,000 to 68,000,000 ha at the beginning of the century, representing 20% of the African forest cover (Sommer 1976). We used the vegetation maps of Chevalier (1912) and of Shantz and Marbut (1923) to assess the forest cover at the turn of the century, remaining cautious in the interpretation of their results.

The second scale is the national level (Figure 2.3). This allows a more precise vision of deforestation in Côte d'Ivoire. The cartographic data used for this approach are the vegetation map of Guillaumet and Adjanohoun (1969) at 1:500,000, which is based on the interpretation of aerial photographs taken from 1955-58, and the NOAA-AVHRR images of 1993 (Dao 1999).

The third scale is the regional level. We chose the region including Abidjan in southeast Côte d'Ivoire, as it is the only region for which a time series of images exists. LANDSAT images taken in 1990 and 2000 cover the region between the coast of the Gulf of Guinea, including Abidjan, and the V-Baoulé (Figure 2.4). This represents an area of 180 X 180 km, about a third of the Ivorian forest zone. This area shows many types of forests and also many types of farming activity, essentially due to the variability in climate and soils. We have also used the topographical maps of the Côte d'Ivoire Geographical Institute (Dao 1999) which are based on aerial photographs taken in 1958 and which are more precise than Guillaumet and Adjanohoun's maps (1969).

The fourth scale is the local level. At this level we analyse changes in forest cover of eight blocks of 20 X 20 km extracted from the LANDSAT images. Both the evergreen forest zone and the moist semi-deciduous forest zone are present here.

The characteristics of the maps and images are summarised in Table 2.1. Treatment of the 1990 and 2000 LANDSAT images was done with IDRISI software. Digitisation of the old maps was done with ARCINFO software, and the graphic production of the maps was done with ARCVIEW. These data were all connected to SIGIVOIRE (Chatelain & Gautier 2002).

Image classification

The vegetation map of 1990 was obtained by classifying three infrared channels of LANDSAT, using 45 classes (the visible channels are unusable). The classes that represented identical or similar subjects were then merged to obtain ten classes. The 2000 map, on the other hand, was obtained by automatic classification with the IDRISI "isocluster" method, which uses 32 classes. These classes were then regrouped to arrive at 15 classes that are identical to those of the 1990 map. Verification of the classification was done by utilising the "kappa" index (Congalton & Mead 1983). We did not keep the degraded forest classification because almost all the forests in this region are degraded, making it impossible to distinguish the states of degradation on the 2000 image, as opposed to that of 1990.

The retained classes are: Forest on firm land: this groups most of the forests (degraded forests, wetland forests, old secondary forests, etc.).

Secondary forest: woody vegetation that reaches 10 m in height, with occasional presence of palms. This includes plantations of cola-nut trees in this region as well as degraded forests with a cover of less than 10%.

Table 2.1 Cartographic data used in the present study.

Year

Scale

Type

1958

1:200,000

Topographical map (IGCI), used for the region around Abidjan. For explanation of the legends, see Dao (1999)

1958

1:500,000

Vegetation map (Guillaumet & Adjanohoun 1969), used for the national approach, based on aerial photos from 1955-58

1990

resolution 30 m

LANDSAT TM (196-56, December 24th 1990)

1993

resolution 1 km

3 images NOAA AVHRR in 1993

2000

resolution 30 m

LANDSAT ETM+ (196-56, February 9th 2000)

VD VD

CO CA CS

CA CA VO

VO CO VA CO

es va m

VO VO CA CA es

CS VA CA VO

VO K

00 VO

CA VO VO VA VO CO CA

VACAVOCAcA-cfVAVO

CA VA VA VA

VA VA CA CS

VO VA CS VO CA VA

Thicket: woody vegetation 1 to 3 m in height. It can be confused with coffee and cacao plantations which are of a similar density and height to thickets. Thickets precede the first stages of secondary forest.

Coffee-cacao: plantations of coffee and cacao, with or without forest cover.

Cultivated grasslands: fields under annual cultivation and grassy fallow lands. This covers large areas of lowlands where rice is cultivated.

Bare ground, village, roads: this is made up of areas with no vegetation.

Orchard plantations: this is made up largely of plantations of oil palm trees, pineapple and banana.

Rubber and other tree plantations: we retained this class because of the existence of large plantations in the North zone and the Dabou region. It is difficult to distinguish this classification from the previous one.

Savanna: we have merged all types of savanna into this classification.

Water: this includes the ocean, lagoons, lakes and rivers.

Before being superimposed for comparison, the 1990 image needed to be geographically corrected based on the topographic maps. The 2000 image was corrected based on the 1990 image. The map of the forest cover that was made based on the 1955 and 1958 aerial photographs could not be superimposed exactly on the satellite images due to the differences in scale.

The identification of palm plantations posed a problem because they were sometimes confused with forest: however, their homogeneous structure and the presence of rectangular trails makes them visible to the naked eye. To deal with them, we manually mapped the plantations and assigned their zones to the corresponding class. The same errors occurred for coffee-cacao plantations, and for the division between secondary forests and very degraded forests. These were of minor importance, however.

Verification of the 2000 map was done on the basis of that of 1990 for which we had already done verifications between 1990-93 within the framework of preceding studies (Chatelain 1996, Dao 1999).

Comparison of images

The 1990 map was qualitatively compared (cross tabulation method) with the 2000 map to obtain 10 X 10 classes (100), then filtered to eliminate isolated pixels. Only 15 classes had a real relevance concerning the development of the forest (Table 2.2). The rest of the classes were merged either because they covered small areas without relevance (e.g. bare ground, water, savanna-grasslands, coffee-cacao thicket etc.) or because they remained ambiguous and could not be interpreted (palm-ground, palm forest). The values of the forest area of 1990 and 2000 and the evolution between 1990 and 2000 are the outcomes of this analysis, so only the forest classification was retained to express the area of forest cover. The secondary forest classification was not kept except for certain statistics, and this will be noted when applicable. Verification of such a comparison was done in a quantitative way with the infrared channel (TM7) of the 1990 and 2000 images. This, after the calibration of the two images, enabled us to find the pixels of which the values have been either increased or decreased following the increase in biomass or reforestation. The superposition of the results of this quantitative comparison with the qualitative approach gives a tool for verification (in the case of the deforested zones we found a concurrence of 70%).

Fragmentation

On the national scale, the pixels on the maps were reclassified into forest and non-forest with the aim to analyse the distance of a forest pixel to the nearest forest limit. For each pixel, the Euclidean distances to the forest limits were calculated. The negative values give the distance from a pixel inside a forested area to the forest limit. The more negative the value the further the distance to the forest limit and the less the external influence on the forest ecosystem. We calculated a "core area" defined as that part of the forest that was more than 1 km away from its border. The positive values give the distance between the blocks of forest. For each block of forest, we calculated a perimeter-area relationship (PA). The PA has a value between 1 (maximum fragmentation, in the case of an isolated pixel) and 0 (theoretical case of an infinite area). The PA index takes into account both the form and the size of the blocks: a square of 4 pixels has a higher index (0.5) than a square of 16 pixels (0.25), but a lower one than a line of 4 pixels (0.62). These methods have been tested and used by Dao (1999).

Results

1. Scale of West Africa

As difficult as it is to get good estimates of recent forest cover with the tools of satellite imagery, as is shown with the disparity of values in Table 2.3, it is even more difficult to get a good view of the forest cover at the turn of the last century or before. The difficulty is even greater since the terms initial, original or primary forest are all relative considering the dynamism of the vegetation over time (Richards 1996). Fairhead and Leach (1998) showed convincingly that the forest cover in West Africa at the turn of the century was over-estimated. We took this into account in adopting the most precise estimates possible by using the original vegetation maps of Chevalier (1920) and Shantz & Marbut (1923) (Figure 2.2B). Although in Sierra Leone the forest mosaic in the forest savanna transition zone indeed was mapped as forest (and thus gave rise to over-estimation of forest - Fairhead & Leach 1998), this was not the case in the other countries in West Africa. Also the coastal zone, which also is a mosaic, was not mapped as forest along its entire length. The forests along rivers, so characteristic of Guinean savanna, were mapped separately, but the size of these areas was indeed excessively large, for typographical reasons. Excluding mosaic zones or forest gallery, Chevalier's 1912 map (Chevalier 1920) shows an area of 29,205,000 ha in Côte d'Ivoire, Liberia, Ghana and Sierra Leone (Table 2.3). The same analysis of Shantz's map (Shantz & Marbut 1923), a less precise map, shows an area of around 20,849,000 ha.

Table 2.3 Forest surface in thousands of ha for different countries and different sources. According to these data, deforestation occurs only in a few countries. (1) Chevalier 1920; (2) FAO/UNEP 1981; (3) Paivinen et al. 1989; (4) SOFO 1997; (5) Odoom 2000; (6) Iremonger et al. 1997; (7): FAO 2000. Countries are ordered from west to east.

Table 2.3 Forest surface in thousands of ha for different countries and different sources. According to these data, deforestation occurs only in a few countries. (1) Chevalier 1920; (2) FAO/UNEP 1981; (3) Paivinen et al. 1989; (4) SOFO 1997; (5) Odoom 2000; (6) Iremonger et al. 1997; (7): FAO 2000. Countries are ordered from west to east.

Country

Total surface

% of

1912

1980

1989

1990

1990

1995

1996

2000

area of the

protected

(1)

(2)

(3)

(4)

(5)

(4)

(6)

(7)

country

forest

Senegal

19200

12.0

2200

210

7600

7300

11100

Gambia

1000

3.7

60

50

100

91

480

Guinea Bissau

3600

0.0

660

690

23.6

2300

2010

Guinea

24500

1.8

2000

760

6700

7663

6367

7660

6929

Sierra Leone

7160

5.1

1238

700

510

1520

1895

1309

1360

1055

Liberia

11000

1.5

8465

2000

4120

4600

4639

4507

6320

3481

Côte d'Ivoire

31800

10.0

13672

4500

310

5600

10961

5469

7780

7117

Ghana

24000

3.3

5830

1710

9600

9600

9608

9022

5970

6335

T°g°

5400

3.4

340

1360

1330

1370

1240

420

510

Benin

11300

14.0

4700

4200

4900

2104

4625

2110

2650

Nigeria

91000

4.7

5900

3860

1430

2790

1370

27900

13517

Total

229960

6.0

24770

25670

45740

43460

73110

If the conservation of the "Congolian" forests is less of a problem due to their extent, the conservation of the forests of Upper Guinea is more and more at risk if an effective management strategy is not put into place. Of the rainforests that at the turn of the last century still covered a small part of Sierra Leone (1,238,000 ha), half of Liberia (8,465,000 ha), half of Côte d'Ivoire, a little under half of Ghana (5,830,000 ha), and a little bit of Togo, today there remains only the Taï National Park in Côte d'Ivoire, a few large forests in Liberia, and some forests in Ghana.

The total area of these forests from Guinea to Nigeria totaled 48,783,000 ha in the eighties (FAO 1993) and 41,594,000 ha in the nineties (FAO 2000), that is, around 60% of the forest at the beginning of the century. Aside from the National Parks, these last large forest tracts are degraded by forestry exploitation.

The amount of deforestation and the policies of protection (definition of protected areas and respect for these areas) differ largely between countries (Table 2.3). According to the FAO (2000), between 1990 and 1995 the annual percentage of deforestation varied between 0.56% for Côte d'Ivoire and 3% for Sierra Leone. While in Ghana the protected areas are part of a forest network, in Liberia the situation is hazier. The most recent forest cover data are those of Paivinen et al. (1992) (Figure 2.2), De Monza (1992) and Parren & de Graaf (1995). The data of the FAO (1993, 1995a, 2000) are simply extrapolations taken from these data, and there is no recent map of the total forest cover in this region. In the case of Côte d'Ivoire, the data of Odoom (2000) and the FAO (1993) showed an increase in excess of 1,500,000 ha of forested area between 1990 and 2000. This, as we will see, is a flawed result, whether or not we take forest plantations into consideration.

2. National scale: Côte d'Ivoire

1,994,100 ha (of which 1,920,700 ha is within the forest zone, the rest being isolated forest patches in the savanna zone). It means that only 6% of the forest part of Côte d'Ivoire is considered as forest zone. These values differ greatly from those published by the FAO (2000).

Thirty-six per cent of the forests are in two administrative regions (Table 2.4) alone - the Southwest and the South - and consist of six main forest areas: the Taï region with the Taï National Park, the Goin-Débé and Cavally "forêts classées"; the Agboville-Abengourou region with the forests of Mabi-Yaya and Yapo, and the Guitry region with the Gô-Bodiénou "forêt classée". At the same time some regions have no forest at all. This is the case most notably in the Daloa-Gagnoa-Soubré triangle in the Central-West region, which was the first zone of agricultural colonisation that moved towards the country's southwest during the 1960s.

Compared with Monnier's (1983) ecological zones, forest is found mostly in the evergreen forest zone (25% of the forest, a total of 1,657,100 ha) and is least present in the moist semi-deciduous forest zone (3% of the forest, a total of 263,600 ha) (Table 2.4).

The protected areas

The total area of protected forests covers 5,667,405 ha in 1992, which represents 17.6% of the area of Côte d'Ivoire, which is considerable (Table 2.4). The "forêts classées" cover 2/3 of the protected areas, while national parks represent 1/3. The "forêts classées" are located mostly in the forest zone, as whilst only 40% of the area of the national parks are located in forest zone. Two national parks alone, Comoé (in the northeast of the country, 1,167,167 ha) and Taï' (southwest, 423,942 ha) total 1,591,109 ha, that is, 81% of the total area of national parks, and 28% of the total protected area. 134 of the 189 "forêts classées" (71%) have an area equal to or less than

The oldest map showing forest cover of Côte d'Ivoire was the one of Chevalier (1920) (Figure 2.2B), which enabled us to estimate the forest cover at approximately 13,670,00 ha in 1912. This value is, in our opinion, indicative, since Chevalier's map did not consider mosaic savanna forests or the entire coastal zone, which he considered non-forest. In contrast, the mosaic zones in community forest gardens and the areas of towns and villages are indistinguishable from the forest. The map of Guillaumet and Adjanohoun (1969) (Figure 2.3), based on data of 1955-58, shows a forested area of 8,640,000 ha. From 1900 to 1955, the forest cover did not change much compared to later years, despite the creation of numerous pioneering fronts of forest exploitation during that period (Kolibi 1990).

In the whole of Côte d'Ivoire, between 1958 and 1993, more than 80% of the rainforest disappeared - about 60,000 km2. According to the 1993 NOAA classification (Dao 1999) (Figure 2.4), the total forest area dropped to

Table 2.4 Forest cover as percentage of total area in the different administrative zones of Côte d'Ivoire in 1993. The forest cover is based on NOAA satellite images. The protected forest percentage is the percentage of the forest in the region.

Region

Forest

% of total

Protected

NOAA (ha)

of the zone

forest (%)

Center

3,400

0

7.9

Center East

49,400

7

14.6

Center North

600

0

15.7

Center West

79,700

3

24.4

North

0

0

6.5

Northeast

44,400

1

28.5

Northwest

100

0

16.3

West

746,200

25

20.9

South

427,000

11

36.5

Southwest

643,400

25

14.7

Total

1,994,100

6

17.6

20,000 ha, and have a cumulative area of only 16% (878,804 ha) of the total 5,667,405 ha.

It is to be noted that the savanna zone contains proportionally more small sized "forêts classées" (less than 10,000 ha), than the forest zone, that is, 66% compared to 43%.

The actual forest cover

The different regions in Côte d'Ivoire have a highly variable area of protected forest cover. The West region of Côte d'Ivoire (28.5% of the area is forest) and especially the Southwest region (36.5%) are most densely covered with protected forests. This high value is principally attributed to the presence of the Taï forest. In the same way, the Northeast region, which includes the Comoé national park has 24.4%. At the scale of Côte d'Ivoire, the regions most poorly covered with protected forests are found in the borders of the North (around Tingréla, in the

Table 2.5 Forest cover of some protected forests. Calculations are based on the NOAA images of1992 and 1993. The fragmentation index Perimeter / Area (PA) and the core index (distance to the interior of the forest) are shown as well.

Name

Classified surface

Forest surface 1992/93 (ha)

Forest surface 1992/93 (%)

Core index area 1993 (%)

PA index (%)

Taï (P.N.)

422,076

393,287

93

88

0.03

Hte Dodo

243,438

160,901

66

44

0.07

Cavally-Goin-Débé

198,473

137,846

69

55

0.06

Mabi-N'To-Songan-Tamin-Yaya

184,689

111,618

60

36

0.10

Ht Sassandra

102,347

69,411

68

55

0.06

N'Zo (R.F.)

76,120

S9,164

78

64

0.09

Niégré

97,468

S7

37

0.07

Scio

90,271

60

39

0.09

Duékoué

Sl^

30^3

S8

38

0.11

Mt-Péko (P.N.)

30,98S

23^6

74

42

0.12

Okromodou-Diogoro-Bogbo

90,149

19,274

21

11

0.06

Go-Bodiénou

61,238

18,176

30

16

0.07

Yapo

2S,373

18^4

71

41

0.10

Tiapleu

19,274

17,932

93

56

0.20

Niouniourou

18,786

17,688

94

69

0.11

Mt Momi-Sangouiné

32,693

^,614

48

10

0.23

Béki-Bossématié

37^72

^,370

41

24

0.06

Asagny (P.N.)

19,396

12,199

63

28

0.19

Mt Glo

10,613

8,29S

78

47

0.19

Krozialé

9,149

7,807

8S

56

0.12

Dogodou

6,343

3,294

S2

25

0.13

N'Guéchié

3^38

3^0

86

38

0.^

Banco (P.N.)

2,196

2,196

100

44

0.21

Mt Bableu

4,148

2,196

S3

26

0.16

Mt Glas

1,464

732

S0

8

0.^

Table 2.6 Number of forest islands per size class (ha) for each of eight 20 X 20 km blocks in the Abidjan region in the year 2000. The percentage of the forest surface of some classes is given between brackets.

Zones

0.3 -2 ha

2-4 ha

4-8 ha

8-16 ha

16-32 ha

32-64 ha

> 64 ha

Total (ha)

Divo

13S

8

8

2

1

1

1

S60

Hiré

34

2

1

0

0

0

0

41

9S2

141

72

31

16

6

S

679S

Yapo

S81

S0

36

lS

S

1

2

12986

Abié

13S8

229

13S

S3

18

S

3

389S

Sikensi

1034 (43%)

127 (21%)

S0 (17%)

14 (8%)

3 (4%)

0

0

1664

Loviguié

1099 (18%)

189 (12%)

81 (10%)

33 (8^%)

18 (9%)

11 (11%)

S (28%)

4286

Dogodou

S77 (8^%)

68 (4%)

46 Ö0/»)

29 (7%)

9 (4%)

12 (12%)

3 (S7%)

4636

Mean

721.2

101.7

S3.6

22.1

8.7

4^

2.3

43S7

Figure 2.5 Area of closed forests in Côte d'Ivoire (1918-90). (Sources: Monnier 1983, Arnaud & Sournia 1978, FAO/UNEP 1981, FAO 1993).

savanna zone) and in the Central-East (around Bondoukou, at the limit of the savanna forest).

In the forest zone, the extreme Southeast and most of the Daloa-Gagnoa-Soubré triangle in the Southwest are farthest from all protected forests. The latter zone was the "frontline" of the colonisation of the southwest since the end of the 1960s: a zone of occupation and movement towards the open spaces of the southwest. It is still subject to intense agricultural exploitation today. To a lesser degree, the Bongouanou-Daoukro-Bocanda triangle (known as the "cacao ring") also has a deficit in forests. The same applies to the area around Adzopé (in the midst of the oldest area of coffee and cacao cultivation) and to the coastal regions of Jacqueville and Bonoua, zones of large industrial plantations, notably of palms and bananas.

Development of protected areas

In general, while the number of protected forests has diminished since 1949, their area has increased. This is the case both for the "forêts classées" and for the national parks. The savanna zone shows an increase in both number and area of "forêts classées". In the forest zone, including the national parks, the highest losses of protected forests were in the Central-West and Central-East regions, that is, on the border of the Baoulé people's original homeland. The largest increase in protected forest area was in the Southwest region, where human influence did not increase much since 1949.

In the "forêts classées" of the forest zone, remnant forests represent on average 42% of their area. The area of these "forêts classées" cover 1,218,000 ha, which represent 68% of the total 1,994,000 ha of forests in the zone. This average is misleading because only 17 forests out of 169 have a real forest cover more than 94% (Table 2.5). Most of them are located in the west of the country, with the exception of the Yapo "forêt classée", the Mabi-Songan-Tamin-Yaya grouping, and Beki-Bossématié.

Fragmentation

An analysis of the 1993 NOAA image shows that the forest is very fragmented. A total of 486 tracts were recognised. Measuring the level of fragmentation using the "Perimeter-Area" index (PA), based on the form of the fragments, shows that these forests have an average value of 0.17. The four forests with a PA below 0.1 are Ta'i, Goin-Debe-Cavally, Haute-Dodo and Haut-Sassandra (Table 2.5).

The core index represents the distance between the black border and the centre of the fragment. Only 22% of the forest areas are found at more than 1000 m from the edge of the forest, 10% are further than 5000 m. Only 14 fragments represent 75% of the forest cover.

3. Regional scale: Abidjan region

The forest covered approximately 2643 km2, or 8% of the Abidjan region's area (Figure 2.6) in February 2000. The area deforested between 1958 and 1990 was 10,580 km2 while between 1990 and 2000 it was 1530 km2 (Figure 2.6). The area deforested since 1990 seems comparatively small, however more than the half of the remaining forest area has disappeared during this interval and there is hardly any forest left. If we consider the percentage of forest cleared in relation to the total forest area over ten-year periods, the two decades 1970-80 and 1980 - 90 show a rate of 30%, while 1990-2000 shows 20%. This is still considerable, and does not indicate a big change despite the development of a forest management policy.

The values given for the deforested areas only consider the transfer from a forest legend to a thicket or cultivated land legend. More of the 1990 forest area has become secondary forest (Table 2.2), which is actually a degradation. However, image interpretation constraints do not allow us to map evolution from thicket to secondary forest.

Development before 1958

In 1958, the forest cover of the Abidjan region was 27,806 km2. Large forest areas were still numerous and the majority of "forêts classées" were still little touched by agriculture. The map of 1958 showed evidence of deforestation mostly in the east of the region (Adzopé, Agboville, Dabou). The region between Grand-Lahou and Divo was still intact, save for a hole expanding from Divo towards Guitry, clearing centred around the main localities and followed the main axes of communication, as the creation of these routes was linked to the extraction of wood and its transport to the ports (Arnaud & Sournia 1978).

Development 1958-1990

A comparison of the 1958 map of land occupation with the 1990 classification (Figure 2.6B) shows that this period of 32 years had a high deforestation, especially between 1975-1980. Between 1958 and 1990, out of a

Figure 2.6 Changes in forest cover in the region around Abidjan. (A) Changes between 1990 and 2000, based on LANDSAT images with 30 m resolution. The limits of the classified forests are indicated. Only the Yapo classified forest and Banco National Parc (near Abidjan) are completely covered with forest. (B) Changes between 1958 (topographic map) and 1990 (LANDSAT image) with a resolution of250 m. The eight blocks of 20x20km each that are studied in detail (see Figure 2.8) are indicated. A full-colour version of this figure can be found in Appendix 1.

Figure 2.6 Changes in forest cover in the region around Abidjan. (A) Changes between 1990 and 2000, based on LANDSAT images with 30 m resolution. The limits of the classified forests are indicated. Only the Yapo classified forest and Banco National Parc (near Abidjan) are completely covered with forest. (B) Changes between 1958 (topographic map) and 1990 (LANDSAT image) with a resolution of250 m. The eight blocks of 20x20km each that are studied in detail (see Figure 2.8) are indicated. A full-colour version of this figure can be found in Appendix 1.

total zone area of 2,507,982 ha, over 51% (1,274,606 ha) was cleared and only 8% (193,163 ha) was left forested. In 1958, 41% (1,040,213 ha) of the zone was already cut. The 3% (84,900 ha) of land reforested between 1958 and 1990 was very fragmented. On average, 87% of the forested areas of 1958 had disappeared in 1990. Several small forested areas were not mapped as forests in 1958. We are certain that there has been no real reforestation, but that in 1958 an agrarian landscape with a mosaic of wooded fragments grown on the top of slopes was dominant. The zone around Abidjan currently is among the regions with the highest forest cover. The zones that were already fragmented by agriculture in 1958 are the least cut today, and those that were still forested at that time are the most deforested. Contrary to Fairhead and Leach (1998), we think that the old maps strongly underassessed the forest cover of the 1955-58 period because the maps did not take the small and very numerous groves into consideration, for reasons of scale (Chatelain 1996).

The study region is covered with evergreen forest and moist semi-deciduous forest. The moist semi-deciduous forests have the highest deforestation rate and this can be explained above all by the development of coffee plantations, especially in the Tiassale-Divo zone. The evergreen forest zone shows less clearing, and the forests saved are for the largest part situated on sites with difficult access (along the Agneby river, the wetland forest near

Dogodou and the Go-Bodienou forest) and where management policies have been established such as for the forests of Yapo.

Development 1990-2000

In 1990, forests covered 7740 km2. During 19902000 deforestation was 20% (1530 km2) of that area and was mainly found in the evergreen forest zone, specially near Bandama (Figure 2.6A). Very large areas of forest have disappeared in the Go-Bodienou classified forest where deforestation began a few months after the image was taken in 1990. There is actually only 50% of this forest remaining. The Agboville zone, with a long history of human occupation, has paradoxically changed little and still has a significant forest area.

The development map (Figure 2.6), shows hardly any large forested areas aside from those of the Yapo classified forest, and to a lesser extent, Banco national park, Azagny national park, the Agneby forest and the Dogodou classified forest. The forests of Mopri and Tene are nevertheless visible because of a small intact forest that is still present.

Condition in 2000

Between 1958 and 1990, half of the forest cover disappeared, and in the following ten years half of the remaining forest was destroyed. From 1990 to 2000, about 140 km2 of forest were lost to cultivation and 3200

Figure 2.7 Comparison between two interpretation maps of LANDSAT images with 30 m resolution. (A) Map from the present study, showing all forested areas independent of their size. Also secondary forests are given showing the density of human occupation. (B) Map from the 1993 "Bilan Forestier" SODEFOR (1993) based on a 1990 image and taking only forest fragments of more than 10,000 ha. The total area is 55x55km. A full-colour version of this figure can be found in Appendix 1.

Figure 2.7 Comparison between two interpretation maps of LANDSAT images with 30 m resolution. (A) Map from the present study, showing all forested areas independent of their size. Also secondary forests are given showing the density of human occupation. (B) Map from the 1993 "Bilan Forestier" SODEFOR (1993) based on a 1990 image and taking only forest fragments of more than 10,000 ha. The total area is 55x55km. A full-colour version of this figure can be found in Appendix 1.

km2 were exploited to such a degree that in 2000 they were identified as highly degraded forest or secondary forest. Secondary forests, mostly in the east of the region, or neighbouring highly degraded forests cover 11,192 km2 (Table 2.2).

Only 23% of the residual forest cover is within the state-controlled "forêts classées". There is almost no information available for 76% of the forest cover, mainly consisting of areas less than 1000 ha in size and not mapped on the Ivorian forestry map (Figure 2.7).

In some cases, like at Divo, it is even possible to delimit classified forest "in negative": more forest exists outside than inside the classified forest. In the case of the Mopri classified forest, the managers have conserved a tiny grove of primary forest in the middle of the forest, which represents at most 10% of the original forest. The rest is occupied by young tree plantations. Irobo classified forest shows the same scenario. Thus, in this region, these two tiny areas represent the last moist semi-deciduous forests.

4. Local scale

We chose to analyse forest cover and changes therein using eight blocks of 20 x 20 km (Figure 2.8 and Table 2.2). The distribution of the forested areas is heterogeneous: some blocks have maintained a forested character despite their proximity to urban centres (Agboville region), while some isolated blocks have long ago lost their forests (Hiré region). Some "forêts classées" in the studied blocks are devoid of all forest. Although a large proportion of forest is still found in the form of large forest tracts (Yapo forest, Go-Bodienou forest), almost the majority of forests are made up of tiny fragments distributed throughout the rural landscape. Their presence certainly plays a significant role in the process of environmental conservation. The many recent clearings indicate a highly dynamic pattern of deforestation in the studied blocks.

The fragmentation of the forest can be assessed by surveying areas on the basis of surface classes (Table 2.6). The fragments of less than 4 ha in total represent between 10 and 30% of the total forested area, which is far from negligible and justifies our opinion that these areas should be taken into account in studies of forest cover.

Discussion

1. Deforestation at various scales

West African scale

West Africa is widely recognised as highly significant for the biodiversity conservation of the African dense forests (Conservation International 2001). The rate of deforestation is considerable, but information on forest cover differs widely depending on the sources (Matthews 2001), some of which show an increase of forest in certain countries (Table 2.2). Fairhead and Leach (1998) showed that the extent of the forest area at the turn of the 19 th century was over-estimated. This was especially the case for

The Hiré block, in the moist semi-deciduous forest zone. A high reduction of forest cover f-om 368 km2 in 1958 to 18 km2 (less than 5%) in 1990, and c. 0.4 km2 in 2000. Two thirds of the forests are less than 2 ha in area, and the "forêts classées" were completely cleared in 1990. In 2000 the whole area is characterised by old coffee-cacao plantations.

The Divo block. Deforestation here is quite similar to Hiré (from 337 to 40 to 6 km2). Forests here are a little more abundant due to numerous steep slopes covered with granite outcrops that are less conducive to agriculture. There is a unique tract of forest which is privately owned (ex IRCC). Partial exploitation of it, unfortunately, started in 1991. The Boubo "forêt classée" was completely transformed in oil palm plantations.

The Gô-Bodiénou block. Forest cover reduced from 196 km2 in 1958 to 227 km2 in 1990 to 67 km2 in 2000. This block has much closed broadleaf forest and was highly isolated until 1990. A large part of the block is occupied by the Gô-Bodiénou "forêt classée". The landscape to the north of this forest is made up of numerous groves. Para-rubber tree and palm plantations have increased heavily. This "forêt classée", along with those of Yapo and Niegré, are the three last large areas of forest in the South zone of the country.

The Lovigué block. Forest cover reduced from 227 km2 in 1958 to 194 km2 in 1990 to 42 km2 in 2000. Most of the forest area is found outside the "forêts classées". In 1969, deforestation already reached the borders of the "forêts classées". In 1990, over 40% of the "forêts classées " of Mafé and Lovigué were already cleared, and in the decade to 2000, most clearing was done in the Mafé "forêt classée". This block, along with that of Abié, shows the highest forest cover in 2000, mostly situated in the rural domain.

The Yapo block. Forest cover changed from 152 km2 in 1958 to 224 km2 in 1990 to 129 km2in 2000. 14,900 ha of forest is situated within the "forêt classée" whose conservation status seems superficially good. However, the forests are extremely degraded due to high logging levels. This block show the lowest rate of deforestation. In spite of the strong human pressure due to the presence of large banana and pineapple plantations south of the forest, numerous small groves of secondary forests are present where cola-nut trees are often abundant. The block shows a coexistence of industrial along with traditional land use.

The Abié block north of the Yapo "forêt classée". Forest cover changed from 129 km2 in 1958 to 139 km2 in 1990 to 39 km2 in 2000. This block was considered completely cleared on the 1958 maps. It is, however, the block with the highest forest cover in the rural areas. Land occupation according to the relief is clearly visible on the images: forests are found on the tops of slopes and rice and taro fields on low-lying lands. The deforested parts are small in area and scattered. This block illustrates that it is meaningless to consider only large areas of forest in the national inventory. This block is the most forested one both in area and in number of forest fragments.

The Sikensi block north of Dabou. Forest cover changed from 182 km2 in 1958 to 149 km2 in 1990 to 17 km2 in 2000. Forests having an area less than 4 ha represent 64% of the forest area. This block was crossed by a road a long time ago and this resulted in the establishment of many farms. In addition, the most important oil palm plantations are located in the south of this region. Clearing in this block is high and large areas of cleared land belong to a single landholder. Many areas far from the main road were only recently deforested.

The Dogodou block. Forest cover changed from 210 km2 in 1958 to 139 km2 in 1990 to 46 km2 in 2000. Numerous migrants have established themselves along the coastal Abidjan-San Pedro road since it was improved in 1992. Coffee-cacao plantations, however, have been established to the northwest of this main road for a long time. The Yocoboue-Tiegba main road also crosses the block and led to agricultural occupation. Deforestation is essentially in the interior of "forêts classées" (10% of the classified forest area). The zone along the road to Tiegba is a mosaic of forest and agricultural land.

Figure 2.8 Changes in forest cover at a local scale for eight blocks of20 X 20 km each in the region of Abidjan (see Figure 2.6).

(A) Blocks located in the eastern part of the area. Left the changes between 1958 and 1990, right the changes between 1990 and 2000.

(B) Blocks located in the western part of the area. Left the changes between 1959 and 1990, right the changes between 1990 and 2000 (for legends see Figure 2.6). A full-colour version of these figures can be found in Appendix 1.

the savanna-woodland zones, most probably due to the incorrect interpretation of pioneer bush vegetation and of secondary forest. These were often interpreted as a forest degradation phase instead of a forest colonisation phase of the savanna (Spichiger 1975, Blanc-Pamard & Spichiger 1973, Gautier 1992b).

Another explanation of this variability is the definition of rainforest, which varies according to each report (e.g. some take plantations into account, others do not). Another aspect is the absence of recent data, since the last maps made on the scale of West Africa date from 1990. This lack of sound information makes it very difficult to evaluate the level of deforestation, which should be a basis for proper management of forest areas. For the conservation of the last few forest areas, it is certainly necessary to redefine protected areas in Liberia and to reassign the roles of the "forêts classées" in Côte d'Ivoire and Ghana.

The actual assessment of the distribution of the forests should also support the numerous confirmations and explanations of climate changes in West Africa (Anhuf 1995, L'Hote & Mahé 1995).

In Côte d'Ivoire, deforestation has been higher in the moist semi-deciduous forest zone than in the evergreen forest zone. The impact of this is aggravated by the higher sensitivity of degraded moist semi-deciduous forests to fire hazard and infestation by pest plants such as Chromolaena odorata. Although the distinction between the two forest zones is fundamental for proper forest management, it is generally absent on the maps of West Africa such as those of White (1983) or of WWF (EcoRegions from Olson & Dinerstein 1998). This seems equally valid for Ghana where the moist semi-deciduous forest zone represents around 70% of the area but accommodates almost the same quantity of forest as the evergreen forests' zone.

Côte d'Ivoire: national scale

Instead of presenting the statistics on the country as a whole, we have tried to show the values of deforestation according to the domains of vegetation and administrative regions, in order to present evidence on the distribution of the last forests. Forests are mostly situated in the southwest of the country (91% of the forest area) and in the evergreen forest zone (86% of the forest area). The forests in the moist semi-deciduous zone cover less and less area, even inside "forêts classées". This is due to large tree plantations within their borders and to the importance of coffee-cacao plantations in the rural domain.

Since 1926, when the French colonial administration started to determine forest areas reserved for forestry exploitation under the form of "forêts classées", it has not been simple to follow the development of these areas. In fact "forêts classées" appear or disappear from various available lists with no official public notice of declassification (Dao 1999).

Of the two categories defined by the 1978 forest decree, the rural forest domain (the area outside the

"forêts classées") has without doubt suffered the highest deforestation. This is mainly the result of the absence of a legal status of these forests. Not all forests in that domain have been cut however. A certain number of forests which are not "forêts classées" still exist today, often as groves that are difficult to access or are unexploitable. In and around "forêts classées" the regulations against clearing for agricultural use, although in principle absolute, have not always been respected. As a result of this, more than half of the "forêts classées" area has been cleared. Numerous factors determine conservation: the existence of real management and protection projects (as for example in the Taï National Park and the Yapo forest), population pressure around the forests, accessibility and exploitability of the forests. For every forest, this mix of factors is different.

The quality of the forests in Côte d'Ivoire, in terms of fragmentation measured by the sizes of the core areas, is weak as only 22% of the forest area has a distance of over 1000 m to the forest border.

Regional scale

One of the most interesting observations concerns the history of the distribution of deforestation: it runs contrary to the general belief that the longer the history of human occupation is, the more the lands are cleared. In almost all cases, the areas most wooded in 1958 have the lowest level of forest cover in 1990 and 2000, and the areas considered under cultivation in 1958 still have a significant forest cover in 2000.

This phenomenon of differential development of deforestation is illustrated by the comparison of the east and west parts of the Abidjan region: in the western part, between Divo and Grand-Lahou, the forests have been transformed into plantations of coffee-cacao, palm or rubber trees, while the eastern part (Agboville) shows a landscape composed of a mosaic of numerous small fragments of forest (mainly secondary) and crops. We conclude that the rate of deforestation is related directly to the type of land use rather than to the density of the population. The lands situated along the road axes prior to 1958 were and are still occupied by populations that profit from and manage the forests and fallow fields in a sustainable way. In the peripheral zones, which were composed of dense forest in 1969, the lands have been occupied quickly and extensively with an aim of planting cash crops, often by non-indigenous populations who came from the northern regions of the country. The forests of these areas that are still wooded have certainly become depleted, as is the case in Yapo for example. Thus, the preservation of these forests is not due to the absence of exploitation or a weak need for agricultural land, but actually to the sustainable management of the lands.

The percentage of deforestation between 1990 and 2000 varies largely over the administative zones (from 6 to 90%) and the overall statistic for the whole country does not give pertinent information. Overall deforestation is

Figure 2.9 Causes of deforestation in Cote d'lvoire. The scheme is based on the models proposed by Turner et al. (1995).

diffuse and touches even the large forest areas that remain in places with difficult access. Some parts in the South zone, on the other hand, show some very large areas recently deforested in order to plant rubber trees. This is particularly the case in the Gô-Bodiénou "forêt classée". This supports our believe that the role of "forêts classées" as a biological reserve for species is rather weak and that currently their main role is that they are a source of exploitable wood.

At the same level, we have shown that deforestation is much less significant in the Abidjan region than in the Taï region where people prefer to clear large portions of forest to appropriate them even if they will not cultivate them right away.

Local scale

We characterised the changes in the forest cover by analysing the size distribution of the forest areas. The eight blocks of 20 X 20 km under study all showed a significant area occupied by small forest tracts of less than 8 ha in size. These attain 30% of the overall forested area. The assessment of these forest mosaics, which are typical in tropical landscapes, is essential because the presence or absence of such a fragmented forest cover leads us to the mode of land use and to the consequences of the maintenance

Oplan Termites

Oplan Termites

You Might Start Missing Your Termites After Kickin'em Out. After All, They Have Been Your Roommates For Quite A While. Enraged With How The Termites Have Eaten Up Your Antique Furniture? Can't Wait To Have Them Exterminated Completely From The Face Of The Earth? Fret Not. We Will Tell You How To Get Rid Of Them From Your House At Least. If Not From The Face The Earth.

Get My Free Ebook


Responses

  • valerie radford
    Why are evergreen forests unexploitable?
    2 years ago

Post a comment