Bob Scholes, CSIR 2003

African savannasas non-linear systems

Bob ScholesDiv of Water, Environment and Forest Technology, CSIR

PO Box 395, Pretoria 0001 South Africa

bscholes@csir.co.za

Theoretical Topics in Ecological Economics

Trieste, February 2003

Bob Scholes, CSIR 2003

Objectives of this talk

• Show how pervasive non-linearities are in ecosystems, using African savannas as an example

• Demonstrate how they lead to alternate configurations of the system

• Explore how these facts might alter the way we manage our interaction with ecosystems

Bob Scholes, CSIR 2003

Outline of the talk

1. Introduction to African savannas

2. Sources and consequences of nonlinearity

3. [Break for stretch and questions]

4. Dynamics of non-linear systems

5. Discussion

Bob Scholes, CSIR 2003

What is a savanna?Scholes (1998) In: Vegetation of southern Africa, CUP pp 258-277

• Biome co-dominated by trees and grass• [Tropical and subtropical] vegetation types

with a discontinuous cover by trees of at least 2.5 m tall of between 5 and 60%, over a [continuous] grass layer [dominated by C4

grasses]Percent canopy cover

Savannas

Bob Scholes, CSIR 2003

The importance of savannas• World’s largest land biome

– Eighth of world surface, 2/3 of Africa

• Second largest Net Primary Productivity (16.8 x 1015 gC/y) and carbon store

• Large ‘natural’ impact on atmosphere– Fires burn 1/3 to 1/10th per year

• Home to 600 million people – Savannas are main source of food and energy

• Centre of biodiversity – 7000 spp in Africa alone

Bob Scholes, CSIR 2003

African Savannas

Bob Scholes, CSIR 2003

Large mammal biomass in African savannasEast 1984 AfrJ Ecol 22, 245-270; Fritz and Duncan 1993 Nature 364, 292-3

Bob Scholes, CSIR 2003

A tale of two savannas

Nutrient-richFineleafed, thornyMammal herbivoryInfrequent fireBNF widespreadEctomycorrhizaeClayey (smectitic)Young, low surfacesAridHigh population

Nutrient-poorBroadleafed, tannicInsect herbivoryFrequent fireLittle BNFEndomycorrhizaeSandy (kaolinitic)Old, high surfacesMoistLow population

Bob Scholes, CSIR 2003

Types of savannas

Ava

ilab

ility

of n

utri

ent

s to

pla

nts

Availability of water to plants

Arid MoistLow

High

Acacia WoodlandsPanicoidae

CombretaceaeWoodlandsChloridoidae

Mopanewoodlands

Miombo-likeWoodlandsCeasalpinaceae(Dialeae)AndropogonaeB

urs

ura

cea

e &

Ca

ppa

race

ae

Aris

tidae

Forest/grassland

Bob Scholes, CSIR 2003

The digestion threshold

Time of year

Dry DryWetGra

ss N

itrog

en c

onte

nt (

%)

1

2

0.8 to 1

‘Sweet’ grazing

‘Sour’ grazing

digestible

indigestible

Bob Scholes, CSIR 2003

The nitrogen fixation threshold

Plants

Soil

Animals

UreavolatilisationDenitrification

Fire

Biological N fixationPhosphorus > 4ppm

Bob Scholes, CSIR 2003

Consequences of thresholds

• If the forage cannot be digested, it accumulates

• Fire consumes the dry grass

• Nitrogen is lost through the fire– Most returns, some is exported to the oceans

• If there is not enough P for N fixation, the system N declines

• The grass is therefore indigestible

Bob Scholes, CSIR 2003

Why do savannas burn?

Fires in Africa, May-Oct 1989Scholes et al JGR 101, 23677

Infertile savannas and grasslandsVan Wilgen & Scholes 1997 In ‘Fires in African savannas’ ch 3.

Bob Scholes, CSIR 2003

Any questions?

Bob Scholes, CSIR 2003

Non-linearity in production

Bob Scholes, CSIR 2003

Sources of non-linearity1. Geometry of root overlap

Stem diameter

Crown diameter

Root diameter

0

0.2

0.4

0.6

0.8

1

1.2

0 5 10 15 20 25 30

Basal Area (m2/ha)

Fra

ctio

nal

co

ver

ProjTreeCov

RootReach

C:D = 25R:D = 60

Bob Scholes, CSIR 2003

Sources of non-linearity cont..

2. The temporal niche

0.0

0.5

1.0

1.5

2.0

2.5

1 2 3 4 5 6 7 8 9 10 11 12

Month of the year

Lea

f ar

ea in

dex

(m

2/m

2) Tree LAI

Grass LAI

All LAI

wet season wet season

Skukuza30% canopy

Bob Scholes, CSIR 2003

Penman-Monteith equation

E = {s(Rn-G)+cp[e(Tz)-ez]/ra}/[s+(1+rs/ra)]

Evaporation

Net solar radiation

Heat fluxInto soil

Vapour pressure deficit= f(temperature, humidity)

Stomatal resistance=f(species, leaf area)

Aerodynamic resistance=f(wind speed, roughness)

Bob Scholes, CSIR 2003

Penman-Monteith + Phenology

0

100

200

300

400

500

600

0 0.5 1 1.5 2

LAI max,tree

Tra

nsp

irat

ion

(m

m)

Etree mm/y

Egrass mm/y

Bob Scholes, CSIR 2003

Beer’s LawI = I0 e-k*LAI

I = Photosynthetic flux density below canopy (mol/m2/s)I0 = PFD above canopy (mol/m2/s)LAI = Leaf Area Index (m2/m2)K = extinction coefficient

=f(canopy architecture, sun angle)

0

0.2

0.4

0.6

0.8

1

0 1 2 3

Leaf Area Index

Fra

cti

on

of

ligh

t p

en

etr

ati

ng K=0.5

Bob Scholes, CSIR 2003

Consequences of non-linearity in production

1. Once an increase in tree cover begins, the economic viability of ranching deteriorates rapidly

2. The optimal pattern of tree clearance is generally complete removal over part of the landscape, rather than part removal in all the landscape

3. Focus on the least-affected areas first

Bob Scholes, CSIR 2003

Take a break!

Bob Scholes, CSIR 2003

Savanna dynamics

• Around the world, savannas have been observed to change, rapidly and irreversibly, from an ‘open’ grassy state to a ‘closed’ woody state following the imposition of high, fixed stocking with domestic animals

• This can lead to economic failure, since production is based on the grazing system.

Bob Scholes, CSIR 2003

Trees vs grassThe coexistence/competition problem

• Competitive exclusion principle– ‘Complete competitors cannot coexist’

• Woody plants and grasses apparently coexist in time and space in savannas

• No other biome has co-dominance by such dissimilar life-forms

• The tree-grass relationship in savannas is prone to sudden shifts to higher tree cover

Bob Scholes, CSIR 2003

Stable or unstable?Scholes and Archer (1996) Ann Rev Ecol Syst 28:517-44

• Equilibrium models– Niche separation

• Rooting depth

• Phenology

– Balanced competition– Preferential predation

• Predictions– Consistent pattern of

tree biomass in relation to soil and climate

• Disturbance models– Fire– Megaherbivores– People

• Predictions– Variable tree biomass in

space and time, weakly correlated with soil and climate

Bob Scholes, CSIR 2003

Niche separation by depth‘Walter Hypothesis’

Walter (1971) Ecology of Tropical and subtropical vegetation

Predictions:More trees with higher rainfall Less trees with more clay

Grass: surface soil onlyTrees:Surface and deep

Note: complete separationnot necessary

Bob Scholes, CSIR 2003

Graphical isoline analysisWalker et al (1981) J Ecol 69, 473-98

Walker & Noy-Meir (1982) Ecol Studies 42,556-590Noy-Meir Ecol (1982) Studies 42,591-609

Woodiness

Gra

ssin

ess

G=0

W=0

WoodinessG

rass

ines

s

Grassland

Thicket

savanna

Bob Scholes, CSIR 2003

The tree-grass-fire system

PredictionsMore trees on infertile soilMore trees with fire suppression

Trees

Grass

Fire-

+

-

Product of signsis positive, thereforeunstable

+/-

Browsers

Grazers

People

--

+

--

--

--

Bob Scholes, CSIR 2003

The fire trapGambiza et al (2000) Ecol Econ 33, 353-68

Bob Scholes, CSIR 2003

DGVM model results(Sheffield model)

0

5000

10000

15000

20000

25000

200 300 400 500 600 700 800 900

Rainfall (mm)

stem

bio

mas

s (g

C/m

2)

fire

no fire

Bob Scholes, CSIR 2003

Global data synthesis

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

0 50 100 150 200 250 300 350

Growth days

Tree

bas

al ar

ea (m

2/ha)

Physiological limit

' The fire trap'

Bob Scholes, CSIR 2003

Flame height

Supresses

Grass biomass

Flame height = f(grass fuel)

GrazersBrowsers

Rainfall

ReducesSlows growth

seeds

Nutrients

Complex tree-grass model

People

Increase &sustain

Harvest

People

Suppress or ignite

Bob Scholes, CSIR 2003

Managing in the presence of abrupt transitions

• Stay away from the threshold by an amount defined by– Variability in the driving forces– Uncertainty in knowledge– Risk tolerance (including ease of reversion)

• Opportunities to alter the state are infrequent– ‘windows of opportunity’

Bob Scholes, CSIR 2003

BifurcationsSavannas have several quasi-stable states

There are slow dynamics and fast dynamics

Nutrient rich savannaon young surface

Nutrient-poorSavanna onOld surface

Grassy mode

Woody mode

Sustained grazingFire exclusion

Intermittent grazingOccasional fire

Millions of years

Enrichment

Hot-spot

Bob Scholes, CSIR 2003

Take-home lessons

• Non-linearities are a very common feature of ecosystems

• They tend to create alternate modes• The modes have profound consequences

for the economic use of the system • Small management changes can lead to a

rapid, and effectively irreversible mode switch– But most systems are probably quite resilient, most of

the time