Methods

Study area, plot- and data selection

The area covered in this analysis is confined to the forest zone of Sierra Leone, Liberia, Côte d'Ivoire and Ghana. These countries contain most of the forest in Upper Guinea and are relatively well studied.

The data we have used are collected by various organisations. For Sierra Leone we used the data from Small (1953), Savill & Fox (1967) and Davies (1987). For Liberia we used data from the German Forestry Mission to

Liberia (GFML 1967a, 1967b, Sachtler & Hamer 1967a, 1967b, Sachtler 1968) and the Liberian Forest Service. For Côte d'Ivoire we drew largely on data collected by SODEFOR (Clément & Guinaudeau 1973, SODEFOR 1978, 1979), and for some sites on species lists (Kouamé et al. Chapter 5). For Ghana we used the national forest inventory data (Hawthorne & Abu Juam 1995, Hawthorne 1995a, 1996, Wong 1989). In Liberia and Côte d'Ivoire only a number of large timber tree species have been surveyed (in each country c. 60 species), whereas in Ghana all species with a diameter at breast height (dbh) over 5 cm were included.

To be able to compare the different areas we made a list of species that were inventoried in all three countries.

Table 4.1 Species used in the ordination/classification analysis. In some cases several species in a genus are pooled. Ordination scores of the first two axes are given for an ordination based on abundance data, and one on presence/absence data.

Presence/absence Species

Species code

Family

Maximum size

Abundance

(m)

(cm)

Axis 1

Axis 2

Axis 1

Axis 2

Afzelia bella & africana

Afz.spp

Leguminosae

3S

>9C

38.9

36.3

42.7

37.9

Alstonia boonei

Als.boo

Apocynaceae

45

14C

21.6

57.2

34.2

50.1

Amphimas pterocarpoid.es

Amp.pte

Leguminosae

sc

120

41.9

63.5

41.0

55.9

Anopyxis klaineana

Ano.kla

Rhizophoraceae

45

120

56.1

46.1

56.1

58.1

Anthonotha fragrans

Ant.fra

Ceasalpiniaceae

38

120

69.2

33.4

68.5

16.2

Antiaris toxicaria

Ant.tox

Moraceae

51

13C

20.2

56.2

20.9

49.9

Berlinia confusa & occidentalis

Ber.spp

Leguminosae

4C

>1CC

47.3

67.1

47.4

59.1

Canarium schweinfurthii

Can.sch

Burseraceae

SC

1SC

51.8

54.0

52.1

56.9

Ceiba pentandra

Cei.pen

Bombacaceae

6C

2CC

23.4

45.4

31.1

47.8

Celtis adolfo-friderici & mildbraedii & sp.

Cel.spp

Ulmaceae

S4

lCC

6.9

54.3

26.8

42.1

Daniellia ogea & thurifera

Dan.spp

Leguminosae

4S

120

49.5

60.8

50.0

51.2

Distemonanthus benthamianus

Dis.ben

Leguminosae

36

97

32.0

50.6

33.3

40.8

Entandrophragma angolense & candollei & cylindricum & utile

Ent.spp

Meliaceae

6C

2SC

24.5

57.4

20.8

48.5

Erythrophleum ivorense & suaveolens/guianense

Ery.spp

Leguminosae

4C

120

55.1

28.1

53.4

31.1

Funtumia africana

Fun.afr

Apocynaceae

3C

S2

42.3

55.7

34.7

52.6

Gilbertiodendron preussii

Gil.pre

Leguminosae

3S

120

77.1

48.3

74.0

0.0

Guarea cedrata

Gua.ced

Meliaceae

4C

lCC

23.6

66.9

26.4

59.4

Guibourtia ehie

Gui.ehi

Leguminosae

4S

>100

15.1

0.0

21.9

1.6

Heritiera utilis

Her.uti

Sterculiaceae

4S

3CC

66.7

51.6

63.5

46.8

Khaya anthotheca & ivorensis & grandifoliola

Kha.spp

Meliaceae

>SC

>180

20.4

63.6

19.8

55.0

Klainedoxa gabonensis

Kla.gab

Irvingiaceae

4C

120

48.0

59.7

45.6

55.6

Lophira alata

Lop.ala

Ochnaceae

SC

1SC

68.0

41.7

63.5

46.3

Lovoa trichilioides

Lov.tri

Meliaceae

4S

1CC

51.2

54.4

53.1

47.9

Mammea africana

Mam.afr

Guttiferae

4C

1CC

49.1

81.4

49.6

100.0

Milicia excelsa & regia

Mil.exr

Moraceae

>SC

>150

33.4

42.9

37.7

43.6

Nauclea diderrichii

Nau.did

Rubiaceae

SC

150

50.2

47.5

47.6

51.2

Nesogordonia papaverifera

Nes.pap

Sterculiaceae

4S

120

1.6

46.1

0.6

40.5

Parinari excelsa & sp.

Par.spp

Chrysobalanaceae

4S

150

55.3

59.5

49.9

61.1

Petersianthus macrocarpus

Pet.mac

Lecythidaceae

4S

180

33.2

67.3

33.3

58.9

Piptadeniastrum africanum

Pip.afr

Leguminosae

SC

180

40.9

60.1

41.3

52.2

Pycnanthus angolensis

Pyc.ang

Myristicaceae

3S

120

39.4

55.3

40.2

48.8

Rhodognaphalon brevicuspe

Rho.bre

Bombacaceae

4S

120

30.5

62.0

11.3

67.4

Ricinodendron heudelotii

Ric.heu

Euphorbiaceae

3C

112

12.6

55.6

10.2

49.1

Terminalia ivorensis

Ter.ivo

Combretaceae

4S

124

40.7

37.8

43.9

44.0

Terminalia superba

Ter.sup

Combretaceae

4S

150

26.5

42.8

27.2

47.3

Tetraberlinia tubmaniana

Tet.tub

Leguminosae

42

1CC

100.0

61.8

100.0

45.3

Tieghemella heckelii

Tie.hec

Sapotaceae

SS

250

41.4

68.8

46.7

64.9

Triplochiton scleroxylon

Tri.scl

Sterculiaceae

SC

>136

0.0

44.0

0.0

42.3

Turraeanthus africanus

Tur.afr

Meliaceae

3S

1CC

22.2

100.0

33.7

80.9

Zanthoxylum gilletii Zan.gil Rutaceae 33 8C 38.1 39.3 14.3 92.8

Zanthoxylum gilletii Zan.gil Rutaceae 33 8C 38.1 39.3 14.3 92.8

For some genera the species are pooled because in some countries the species were identified to genus only (this is the case for Afzelia, Berlinia, Celtis, Entandrophragma, Erythrophleum, Khaya, Milicia, Parinari). In addition some species were added that occur in one country and are absent in the others (for instance Tetraberlinia tubmaniana is found only in Liberia). In total we have selected 40 species or species groups. In Table 4.1 the selected species are listed along with several characteristics. Most of the species have maximum sizes of over 40 m height and over 100 cm in diameter. Almost all selected species have commercial value, mostly as timber species.

In total we collected species abundance data for 176 forest sites, 8 for Sierra Leone, 26 for Liberia, 37 for Côte d'Ivoire and 105 for Ghana. The inventoried area of each site is variable and ranges from 10 to 4500 ha (Appendix 2). Inventories vary from complete inventories to strip inventories. The 176 sites cover a wide range in environmental conditions (Table 4.2) (see below for sources of data). Rainfall varies between 1200 and 3500 mm per year, water holding capacity between 10 and 115 mm water/m soil, cation availability (Ca2+, Mg2*, K+) between 0.3 and 40 cmol per kg soil. The altitudes of the sites vary between 50 and 760 m above sea level. Latitude varies between 4.8 and 8.9 degrees and longitude between —11.4 and —0.5 degrees.

Additionally, 38 sites were available with small areas (<10 ha), incomplete species abundance data, or only a species presence-absence list (Appendix 2). These sites were excluded from the analyses but included in the final forest classification and forest map.

Classification and ordination of forests

The main vegetation data consist of a species-plot matrix with 40 species and 176 plots with abundance values for species. For the abundance data the estimated number of trees >30 cm dbh per km2 was used. In Ghana all trees >30 cm dbh were inventoried and data are available on individual trees. All Liberian and Sierra Leonean sites had data for trees over 40 cm dbh and for Côte d'Ivoire various lower limits were available (>10 cm, >15 cm, >20 cm, >40 cm and >60 cm). To be able to compare all sites we estimated for each of the sites the number of trees >30 cm dbh per km2 as follows.

Transformation values were based on the Ghanaian

Table 4.2 Values for selected environmental parameters for 176 forest sites in Upper Guinea, rainfall (in mm per year), CMK (cation availability in cmol per kg soil), WHC (water holding capacity in mm water per m soil) and altitude (m above sea level).

J

N

Mean

Std. Dev.

Minimum

Maximum

176

1780

450.6

1194

3422

CMK

176

3.8

9.3

0.3

39.3

WHC

165

68.7

36.3

10

112

Altitude

176

218.5

111.5

51

758

data. For the Ghanaian data, for each species (thus the pooled data for all sites together) we constructed frequency distribution diagrams for size classes. Such a frequency distribution shows an idealised population structure. Most species then show a negative exponential pattern with size. For each species a negative exponential curve was fitted on the frequencies by size class. We used the regression to estimate numbers of trees >30 cm for each of the sites in Sierra Leone, Liberia and Côte d'Ivoire. We thus assumed that in the sites studied the frequency distributions would be similar to the idealised population structure. For the large samples sizes this is a reasonable supposition.

We classified these 176 sites using a hierarchical classification algorithm, using Ward's method for cluster optimisation and squared Euclidean distances. The classification resulted in seven clusters of sites.

The log-transformed abundance matrix was also used for a detrended correspondence analysis, in which we reduced the multidimensional vegetation space into a two dimensional one, using Canoco (Ter Braak & Smilauer 1998). The first axis represents the main variation in species composition, the second axis the main variation once the first axis variation is removed. The analysis leads to scores for each site and for each species on the first two axes. The same procedure was followed for the same sites using presence/absence data. This allowed us to establish a relationship between the axis scores based on abundance and based on presence/absence.

The results of the hierarchical site classification were overlaid on the ordination diagram to help determine a final classification of all sites into seven groups (forest types).

For the 38 additional sites a presence/absence axis score was calculated based on the species presence/absence axis scores. Using the relationship between presence/ absence axis scores and abundance axis scores we calculated abundance axis scores for these 38 sites.

Two forest maps of the Upper Guinea forest zone were constructed, one based on the first ordination axis, and one based on the hierarchical classification of the 176 sites into forest types. Interpolation of scores or classes between the sites was done using ArcView GIS (ESRI Inc.).

Vegetation-environment relationships

The resulting axes scores from the correspondence analysis (abundance values used) were related to environmental factors that are likely to influence large scale vegetation patterns such as total yearly rainfall (mm/yr), soil fertility (CMK, in cmol cations per kg soil), soil water holding capacity (WHC, in mm water/m soil), altitude (m) and geographical position (latitude and longitude). A rainfall map was created based on data from 578 weather stations in the region. Data on soil fertility and water holding capacity were calculated based on the FAO soil map of Africa and a quantitative review of chemical analyses of soil profiles (Batjes 1997). The water

gua.ced pet.mac tie hec ber.all

[email protected] ric.heu ant.tox amp.pte n.gilP- afr klad|Sba" par.all

'car^nafr tefrîàh nau.did cel.pen ter.sup mil.er afMriV

trebi-L-

40 60

axis 1

Ghana Côte d'Ivoire Liberia Sierra Leone gil.pre lop.ala trebi-L-

B

• *

*

/

" ■ o m ■

»fa

0 A 7 ■ * »

AAÏ

■ ■

•»* • A

A

r, O

A 4

A

axis 1

Ghana Côte d'Ivoire Liberia Sierra Leone

Figure 4.2

Scatterplots of the first two axes of a species-site ordination. The ordination is based on log-transformed abundance data (number of individuals >30 cm dbh per km2).

(A) Species. The abbreviations refer to the species in Table 4.1.

(B) Sites. The symbols refer to the country.

(C) Sites. The symbols refer to the hierarchical classification of the sites into seven clusters (see also Figure 4.4). The lines in the figure indicate the final classification of the sites (combined results of classification and ordination).

HW = Hyper Wet; WE = Wet Evergreen (type 1,2, 3); ME = Moist Evergreen; MS = Moist Semi-deciduous; DS = Dry Semi-deciduous; SL = Sierra Leone type (in classification, part of HW).

holding capacity was calculated based on soil depth and soil texture. It was presumed that sandy soils have a water holding capacity of 75 mm/m, loamy soils of 100 mm/m, and clayey soils of 125 mm/m. For a detailed account on methods and analyses of the environmental metadata see Chapter 9. All spatial analyses were carried out using ArcView GIS (ESRI Inc.). To estimate environmental variables for a forest site, we used its centre.

The ordination axis scores were related to each of the environment parameters using a Pearson's correlation. To evaluate which of the environmental factors was most important in determining variation in species composition, the ordination axes 1 and 2 are regressed on the environmental parameters (and their squared valued to account for possible non-linear effects) using a stepwise multiple regression. Similarly, the axes were regressed on the geographical parameters latitude and longitude.

Species responses to environmental factors

The major vegetation gradient should reflect the distribution gradients of the major species involved. We show some examples of species distributions over the area and also their distribution with respect to the most important environmental gradient underlying the vegetation gradient, i.e. rainfall. The species distributions are made in ArcView, based on log transformed abundance data (interpolations and the values for each site). The non-linear species responses were modelled by regressing log-transformed abundance against rainfall and its quadratic term.

Also, for each of the species studied we have analysed the major environmental factors determining its distribution and abundance. We used a stepwise multiple regression with the four environmental factors (rainfall, cation availability, water holding capacity and altitude) and their quadrates as the independents and species abundance as the dependent variable.

Figure 4.3

(A) Spatial interpolation of the site ordination axis 1 scores, indicating the most important gradient in the species composition. For the analysis a selection of 40 species (in some cases species groups) is used. The size of the symbols indicates the axis scores. A large symbol (and dark interpolation colour) indicates a high axis 1 score (wet forests), a small symbol (and light interpolation colour) a dry forest.

(B) Spatial interpolation of rainfall in Upper Guinea, based on 580 weather stations in the area.

FP43I

FP43I

(C) Spatial interpolation of the forest types (based on final classification from Figure 4.2C, the classes between the lines). The symbols indicate the hierarchical classification results (the seven clusters). Sites with crosses are based on transformation from presence/absence ordination scores to abundance axis ordination scores.

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


Post a comment