Kindt Seminar 7 24th November

Potential natural vegetation mapsfor western and central Kenya

Presently underutilized tools for the selection of indigenous tree species and

their seed sources

Roeland Kindt (ICRAF / VVOB [Flanders, Belgium])Jens-Peter B Lillesø (ICRAF / FaL [Denmark])Paulo van Breugel (ecologist, formerly IPGRI)

Overview• Which species to plant in a certain area for a

certain purpose: use of potential natural vegetation maps to indicate ecological suitability and databases/books to select potential functions– How were the maps and species lists developed?– Do potential natural vegetation maps provide an

adequate picture of climatic/edaphic variation in Kenyan highlands?

– Do potential natural vegetation maps delineate assemblages of indigenous tree species?

– What are the options for agroecosystem diversification?

Map of potential natural vegetation of south-western Kenya

• High resolution maps available for south-western Kenya (Trapnell and co-workers)– Four 1: 250,000 maps with vegetation boundaries for 1960

based on aerial photographs (1:30,000 and 1:50,000) and dense traverses (< 1 mile apart where accessible roads)

– Photographs mainly 1945-63 and fieldwork mainly 1945-1963– Vegetation classified in 18 groups, 23 subgroups, 55 classes

and 217 subclasses– Interpretation of climax vegetation types and eco-climatic

conditions through remnants of climax types and pioneer (secondary) vegetation types

– Unfortunately limited documentation of criteria for vegetation types

– Publication of maps took long time (1966-1986)– Maps have not been used in the realm of agroforestry

1966

1969

1976

1986

17 potential natural vegetation types

• Potential natural vegetation = the vegetation structure that would become established if all successional processes were completed under the present or future climatic and edaphic conditions

• Determined from names of original vegetation types and eco-climatic maps• Main classification scheme is physiognomic (based on structure such as

percentage aerial cover and height) (similar in other schemes)• Secondary classification scheme is floristic (based on dominant or typical

species)• Other differences between types are interpretation of climatic conditions, but

not determined from rainfall or altitude criteria (eg dry montane forest)

• Four sheets produced (1:300,000; A3 format)• Excel sheet with uses and vegetation types compiled for 362 tree species

that are indigenous to Kenya (types from legend map, other paper by Trapnell on forests, other literature, herbarium vouchers; uses from AgroforesTree Database + Useful trees book by RELMA)

• Herbarium locations obtained for 110 species (but > 20 for only 2 species)• Detailed documentation of map interpretation and vegetation-specific lists

almost finalized (want to avoid problem with original map!)

Physiognomic vegetation typesForest

touching and interlocking crowns≥ 8 (10) m talllianas

Woodland≥ (40) 50% cover, open≥ 8 m tall

Wooded grassland (savanna)10 – (40) 50% cover≥ 6 m tall

Bushland and thicket (impenetrable)≥ (40) 50% cover3 - 7 m tall

Bushed grassland10 – (40) 50% cover< 6 m tall

Special types: swamp, bamboo, afro-alpine, moorland

Bamboo woodland and thicket

Mountain scrubland and moorland

Afro-alpine

High mountain vegetation types (3)

Mountain scrubland and moorlandAfro-alpine

Bamboo woodland and thicket

Moist intermediate forest

Dry montane forest

Moist montane forest

Dry intermediate forest

Forest vegetation types (4)

Dry montane forestMoist montane forest

Moist intermediate forest Dry intermediate forest

Lowland Acacia woodland, bushland and thicket

Upland Acacia woodland, savanna and bushland

Woodland and savanna vegetation types (4)

Moist Combretum-Terminalia savanna Dry Combretum savanna

Moist Combretum-Terminalia savanna Dry Combretum savanna

Upland Acacia woodland, savanna and bushland Lowland Acacia woodland, bushland and thicket

Lowland Acacia woodland, bushland and thicket

Evergreen and semi-evergreen bushland

Bushland and thicket vegetation types (3)

Semi-evergreen thickets

Evergreen and semi-evergreen bushlandSemi-evergreen thickets

Lowland Acacia woodland, bushland and thicket

Acacia and allied vegetation on soils with impeded drainage

Swamp and Papyrus

Vegetation types on soils with impeded drainage (3)

Open grassland areas on soils with impeded drainage

Open grassland areas on soils with impeded drainage

Swamp and Papyrus

Acacia and allied vegetation on soils with impeded drainage

Distribution of vegetation types, climate and species

• Both vegetation and species distribution can be explained by the same explanatory variables (biotic, abiotic, landscape configuration, evolutionary time)

• Literature on vegetation types lists typical species for each type

• Flora, databases, books and herbarium records list vegetation types (habitat) for each species

• Present and historical climatic conditions can be determined from pollen composition

• Vegetation map of Africa (White) turned out to be map of plant endemism (phytochoria) and was later used as biogeographical map for African terrestrial ecoregions

• Sophisticated statistical models perform better when vegetation (landuse) is one of the explanatory variables

Biome 4 model (Kaplan et al. 2003)Equilibrium distribution of 28 major potential natural vegetation types (biomes) from latitude (photosynthetically active solar radiation), atmospheric CO2 concentration, mean monthly climate (mean monthly precipitation, temperature, and percent sunshine) and soil physical properties (water holding capacity and percolation rate)

Check of relevance of maps

• Vegetation boundaries vs. patterns in climatic/soil differences

• Vegetation boundaries and original species composition vs. current species composition

Interpolated surface layers

Data layer Resolution

Annual precipitation 5 km (grid)

Annual potential evapotranspiration 5 km (grid)

Mean minimum temperature of the coldest month

5 km (grid)

Number of dry months 5 km (grid)

Rootable depth 1:1 000 000 (vector)

Cation exchange capacity 1:1 000 000 (vector)

Soil water pH 1:1 000 000 (vector)

Percentage of clay 1:1 000 000 (vector)

Percentage of sand 1:1 000 000 (vector)

Altitude (DEM) 92 m (grid)

Slope 92 m (grid)

Topographic wetness index 92 m (grid)

• Investigation how well potential vegetation types correspond to climate, soil and DEM information

1000 2000 3000 4000

600

800

1000

1200

1400

1600

1800

Altitude

Pre

cipi

tatio

n

ALP

BAM

DCO

DIF

DMFEB

IAC

LAC

MCOMIF

MIX

MMFMSM

OGRSET

SWA

UAC

Interpolated surface layersAfro-alpine

Montane scrublandand moorland

Bamboo

Crosses indicate 10%-25%-75%-90% quantilesand are centred on mean

Altitude

Dry montane forest

Evergreen bushlandUpland AcaciaLowland Acacia-Commiphora

Pre

cipi

tatio

n

Forests

1500 2000 2500

800

1000

1200

1400

1600

Altitude

Pre

cipi

tatio

n

DIF

DMF

MIFMMF

moist montanemoist intermediate

dry intermediatedry montane

1000 1500 2000 2500 3000

600

800

1000

1200

1400

1600

1800

Altitude

Pre

cipi

tatio

n

DIF

DMF

MIFMMF

Convex hulls delineate all observations; line types show concentric hulls after outer hull was left out

Altitude

Pre

cipi

tatio

n

Woodlands

800 1000 1200 1400 1600 1800

600

700

800

900

1000

1100

Altitude

Pre

cipi

tatio

n

LAC

UAC

Upland Acacialowland Acacia-Commiphora

500 1000 1500 2000

600

800

1000

1200

1400

Altitude

Pre

cipi

tatio

n

LACUAC

Altitude

Pre

cipi

tatio

n

1000 1200 1400 1600 1800 2000 2200 2400

600

800

1000

1200

1400

Altitude

Pre

cipi

tatio

n

DCO

IAC

MCO

MIXUAC

SavannasMoist Combretum-Terminalia

Impeded Acacia

Mixtures of evergreenbushland and broad-leaved savanna

Dry CombretumUpland Acacia

500 1000 1500 2000 2500 3000 3500

600

800

1000

1200

1400

1600

1800

Altitude

Pre

cipi

tatio

n

DCO

IAC

MCO

MIXUAC

Altitude

Pre

cipi

tatio

n

500 1000 1500 2000

600

800

1000

1200

1400

1600

Altitude

Pre

cipi

tatio

n

EBLAC

MIX

SET

UAC

800 1000 1200 1400 1600 1800 2000

600

800

1000

1200

1400

Altitude

Pre

cipi

tatio

n

EBLAC

MIX

SET

UAC

Bushland and thicketSemi-evergreen thickets

Evergreen bushland

lowland Acacia-Commiphora

Altitude

Pre

cipi

tatio

n

Upland Acacia

0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00% 90.00% 100.00%

Dry Montane Forest

Lowland Acacia

Alpine

Mountain moorland

Moist Montane Forest

Semi-evergreen thicket

Evergreen bush

Moist Intermediate Forest

Upland Acacia

Impeded Acacia

Mixtures

Moist Combretum

Bamboo

Dry Combretum

Swamp

Dry Intermediate Forest

Impeded grassland

All

Correct Same physiognomy Wrong

Linear discriminant analysis: predictions based on GIS layers

Linear discriminant analysis: predictions

0 1000 2000 3000 4000 5000 6000 7000

Dry Montane Forest

Evergreen bush

Impeded Acacia

Moist Montane Forest

Lowland Acacia

Moist Intermediate Forest

Semi-evergreen thicket

Moist Combretum

Upland Acacia

Bamboo

Dry Combretum

Mountain moorland

Mixtures

Dry Intermediate Forest

Alpine

Swamp

Impeded grassland

correct physiognomic wrong

Ecotones and/or resolution?

Dry montaneforest

Moist montaneforest

Mount Kenya

Dry montaneforest

Moist montaneforest

Linear discriminant analysis: commissions (false predictions)

0 500 1000 1500 2000 2500

Impeded Acacia

Dry Montane Forest

Lowland Acacia

Evergreen bush

Moist Combretum

Dry Combretum

Upland Acacia

Semi-evergreen thicket

Moist Montane Forest

Moist Intermediate Forest

Bamboo

Mountain moorland

Mixtures

Dry Intermediate Forest

Alpine

Swamp

Impeded grassland

< 1 km 1 - 5 km > 5 km

Floristic differences

-0.5 0.0 0.5

-0.4

-0.2

0.0

0.2

0.4

Dim1

Dim

2

BAM

DCO

DIF

DMF

EB

IAC LAC

MCO

MIF

MMFSETUAC

principal coordinates analysis based on Bray-Curtis distance for presence-absence of 362 species (Biodiversity.R)

-4 -2 0 2 4

-2-1

01

2

CAP1

CA

P2

altitudealtitudeBAM

DCODIF

DMF

EB

IAC LAC

MCOMIF

MMF SET

UAC

precipitation

Floristic differencesdistance-based redundancyanalysis based on Bray-Curtis distance(Biodiversity.R)

Current patterns of indigenous tree diversity around Mount Kenya

Survey by Ogi et al.

250 quadrats of 50 × 100 m2

within map

279 indigenous tree species(174 species also in literature description)

Total and shared species richness between literature and current species assemblages

Potential Natural Vegetation Type

n Species total (literature)

(Based on total shared)

Species confirmed by survey

% Species total (survey)

Kulczynskiecological distance

Moist intermediate forest

57 105 51 31 61% 82 0.41

Dry Combretum 40 23 21 18 86% 108 0.45

Dry montane forest 37 91 58 31 53% 83 0.42

Moist montaneforest

37 99 46 30 65% 85 MIF (dif 0.005)

Lowland Acacia-Commiphora

25 92 48 35 73% 102 0.36

Evergreen bushland 16 44 38 18 47% 52 0.47

Dry intermediate forest

15 74 49 27 55% 63 0.43

Upland Acacia 7 22 20 6 30% 35 ST (dif 0.108)

Acacia (impeded) 6 28 18 5 28% 18 UA (dif 0.042)

Semi-evergreen thicket

6 29 19 6 32% 52 DC (dif 0.233)

Frequencies of speciesPotential natural vegetation type

Species Rank Frequency (%)

Moist intermediate forest Cordia africana 1 67 n = 57 Croton macrostachyus 2 60 Commiphora eminii 3 53 Bridelia micrantha 4 51 Markhamia lutea 5 37 Erythrina abyssinica 6 33 Croton megalocarpus 7 32 Catha edulis 8 23 Prunus africana 9 21 dry Combretum savanna Croton macrostachyus 1 45 n = 40 Combretum molle 2 42 Euphorbia tirucalli 3 40 Combretum collinum 4 35 Croton megalocarpus 5 32 Piliostigma thonningii 5 32 Azanza garckeana 7 30 Senna singueana 8 25 Dry montane forest Croton megalocarpus 1 62 n = 37 Euclea divinorum 2 51 Juniperus procera 3 49 Plectranthus barbatus 4 41 Rhus natalensis 5 38 Lippia javanica 6 35 Scutia myrtina 7 30 Psiadia punctulata 8 27 Solanum incanum 8 27 Olea europaea 10 22

Frequencies of speciesPotential natural vegetation type

Species Rank Frequency (%)

Moist montane forest Croton megalocarpus 1 70 n = 37 Croton macrostachyus 2 49 Species = 85 Commiphora eminii 3 41 S1 = 42 Prunus africana 3 41 Bridelia micrantha 5 27 Clerodendrum johnstonii 6 22 Cordia africana 7 19 Erythrina abyssinica 7 19 Vitex keniensis 7 19 Vangueria infausta 10 16 Lowland Acacia-Commiphora

Acacia tortilis 1 64

n = 25 Terminalia brownii 2 60 Species = 102 Melia volkensii 3 52 S1 = 49 Albizia anthelmintica 4 40 Acacia senegal 5 36 Acacia ataxacantha 6 32 Berchemia discolor 6 32 Acacia mellifera 8 28 Combretum aculeatum 8 28 Senna singueana 8 28 Evergreen bushland Croton megalocarpus 1 63 n = 16 Euclea divinorum 2 44 Species = 52 Ipomoea kituiensis 2 44 S1 = 26 Acacia drepanolobium 4 38 Plectranthus barbatus 4 38 Acacia xanthophloea 6 31

Selection of species for agroecosystem diversification

• Select frequent species? Promote underutilized species? Balance with exotic species?

• Timber, for example– All confirmed species?– Faster growing primary species with relatively high current

frequencies?• Juniperus procera (49 and 27% of quadrats in dry forests)• Vitex keniensis (19 and 18% of quadrats in moist forests)

– Faster growing primary species with low current frequencies?• Hagenia abyssinica (1 quadrat), Zanthoxylum gillettii (none),

Podocarpus latifolius (none)– Slower growing primary species?

• Olea europaea (22%), Ocotea usambarensis (1 quadrat), Cassipourea malosana (2 quadrats), Podocarpus falcatus (none)

Limitations of vegetation maps to map species distribution

• Changes in climatic or soil conditions from those associated with the map– Need for successional processes (pioneer, climax)– Ecosystem restoration: first abiotic, then biotic filters? – Changes in ecological/dispersal/habitat pools? (landuse, climate

change, invasive species)• Mapped vegetation types are often mosaics with some

small vegetation types that differ from the main type• Vegetation types have ecotones where each species

reaches another environmental limit (fuzzy boundaries between vegetation types), whereas maps show hard boundaries (indicate ecotone width by LDA?)

• Species consist of different populations (provenances) that differ in adaptation to local conditions (precautionary principle!)

Seed sources of Calliandra calothyrsus identified during seed source survey (2004)

21

4

11

4126

3530

23

19

2018

6

12

33

15

31

3628

22897 29

24

34

2

5

13

32

4325

27

381716

55

54

5049

4445

4846

47

5352

51

57

56

1

3

424014

39937

Poor seed sources

Good seed sources

Correspondence to other vegetation classification schemes

11a

11a

44

42

4242

42

45

19a19a

45

45

19a

65

45

54

White. 1983.1:5,000,000

Good correspondence for high mountain vegetation(65), montane forest (19a), evergreen and semi-evergreen bushland (45),moist Combretum savanna (11a), lowlandAcacia bushland (42) and semi-evergreenthickets (45)

Correspondence to other vegetation classification schemes

AT1005

AT0108AT0108

AT0108

AT0108

AT0721

AT0711

AT0711

AT0711

AT0716

AT1313

Olson et al. 2001.

1:5,000,000From White

11a

11a

44

42

4242

42

45

19a19a

45

45

19a

65

45

54

What happened with evergreen and semi-evergreen bushland, semi-evergreenthickets and in the western part of the map?

Boundary between lowland Acacia types?

Conclusions• Bad news: good models and detailed maps for species

distribution or suitability require good presence-only or presence-absence data, and detailed input maps for large range of explanatory variables, whereas neither are commonly available for most components of biodiversity (including tree populations!)

• Good news: potential natural vegetation maps can provide a reasonable summary of climate and the potential distribution of indigenous tree species, they are available for most places on earth and information is available on their species assemblages

• Best news: we already compiled information for a couple of hundred species for a detailed map for central and western Kenya + confirmed some of climatic/floristic information + have information for their uses for many

• Way forward: combine existing potential natural vegetation maps with more extensive set of presence-data and GIS layers to build better species suitability maps

A (some) word of thanks

• Meshack• Sammy, Jonathan, Sally-Anne, Walter• Trees and markets (Tony)• Our donors• Everybody in the audience today

Topics for discussion?• How confident should users

be when using the maps or we when we advise?

• Further testing of maps• Expansion of maps to

White/WWF ecoregions, Eastern Africa, …

• Sharing of information (printed maps, website, documentation)

• How to deal with biotic and abiotic changes