Faculty of Geosciences Copernicus Institute of Sustainable Development Environmental Sciences

(1)

Faculty of Geosciences Copernicus Institute of Sustainable Development Environmental Sciences

Layout: C&M – Faculty of Geosciences – ©2015 (8865)

Stefan C. Dekker ¹ , Zun Yin ² , Mara Baudena ¹ , Bart van den Hurk ³ and Henk Dijkstra ²

(1) Utrecht University, Environmental Sciences, Copernicus Institute of Sustainable Development, Utrecht, Netherlands (s.c.dekker@uu.nl), (2) Utrecht University, Institute for Marine and Atmospheric Research,

(3) Royal Netherlands Meteorological Institute de Bilt

Introduction

• Land-atmosphere feedbacks can generate sudden shift in the vegetation state.

• Bimodal distributions of woody cover with mean annual precipitation provide evidence that alternative stable

states may exist 1,2 .

• Understanding clearly the climate conditions behind this bimodality is important to predict crucial transitions due to climate change

Results from Satellite (Modis) data ³

• Bimodality also found with Mean Annual Radiation (Fig.1)

• Cell by cell analyses show only bimodality at boundaries between grassland-savanna and savanna-forest (Fig.2)

• Best prediction of Land Cover is with a combination of Mean Annual Precip, dry season length, and seasonality (Fig. 3)

How do global models represent savannas? ⁴

• Three Dynamic-Global Vegetetation Models are used to model savanna-forest systems

• Ecological theory: grass-fire feedback is able to simulate bimodality

• Tree-grass fire feedback are differently included in the model

Conclusions

• Bimodality is not observed in current data of woody cover and biomass per grid cell

• Bimodality found with precipitation, radiation and other forcings due to strong correlation between forcing data

• Seasonality is important to predict bimodality

• A bimodal systems can be bistable due to the mechanisms at play: Tree-Grass fire feedback and seasonality are

important

References

1. Staver, A. C., Archibald, S. & Levin, S. A. The Global Extent and Determinants of Savanna and Forest as Alternative Biome States. Sci. 334 , 230–232 (2011).

2. Hirota, M., Holmgren, M., Van Nes, E. H. & Scheffer, M. Global Resilience of Tropical Forest and Savanna to Critical Transitions. Sci. 334 , 232–235 (2011).

3. Yin, Z., Dekker, S. C., M. Van Den Hurk, B. J. J. & Dijkstra, H. a. Bimodality of woody cover and biomass across the precipitation gradient in West Africa. Earth Syst. Dyn. 5, 257–270 (2014).

4. Baudena, M, Dekker S.C.. et al. Forests, savannas and grasslands: bridging the knowledge gap between ecology and Dynamic Global Vegetation Models. Biogeosciences Discuss. 11, 9471–9510 (2014). (in press for BG)

Figure 3 Prediction Land-cover

a. Prediction only with mean annual Precipitation (P) b. Prediction with P and length dry season (LD)

c. Prediction with P, LD, Entropy monthly precip d-f. Difference between 3c and 1a

F, S, G are resp Forest, Savanna and Grass.

s means stable, b means bistable.

‹ Figure 1 a) Map of Woody Cover (W) and b. above ground biomass (B). c,d,e Histograms of W, B and mean annual radiation

Figure 2 Observed bimodality in Woody Cover per gridcell Bimodality of S-F(Savanna-Forest) and G-S (Grass-Savanna) only observed at boundaries. Bimodality criterion with

Integrated Completed likelyhood

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(a)

0.2 0.4 0.6 0.8 W (−)

lat.l

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(b)

2 4 6 8 10 12 14 B (kgC m⁻²)

fcr

0.0 0.2 0.4 0.6 0.8 1.0

0123

4 (c): W (−)

bior

0 5 10 15

0.000.050.100.15 (d): B (kgC m⁻²)

180 200 220 240 260

0.0000.0150.030

(e): R (W m⁻²)

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

B G G−S S S−F F

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(a)

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(b)

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(c)

B G

s

G

b

S

b

S

s

S

b

F

b

F

s

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(d):Forest

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(e):Savanna

20°W 10°W 0°E 10°E 20°E 30°E

5°S0°N5°N10°N15°N20°N

(f):Grass

−

= +

Figure 4 Model results of JSBach, LPJ-Guess, aDGVM.

All models show water-limitation. Due to negative grass- fire feedbacks, only aDGVM show bistability.

Figure 5 Positve grass-fire feedback (a) explaining observed bimodality in Modis tree cover data (b).

0 20 40 60 80 100

0 500 1000 1500 2000 2500 3000

tree cover (%)tree cover (%)tree cover (%)

mean annual rainfall (mmy⁻¹) A

B

C

grass as fuel

competition tree-grass (light and water)

Grass

A

Savanna Trees

-

Ecological theory & aDGVM

-

- +

Forest Trees

water competition grass-tree seedlings

-

- -

Light competition

damage to trees

grass as fuel

competition tree-grass (light and water)

Grass

C

Savanna Trees

-

LPJ-GUESS-SPITFIRE

-

- +

Forest Trees

water competition grass-tree seedlings

-

- competition ^Light

damage to tree seedlings grass as fuel

Competitive exclusion of trees on grass (light)

Grass

B

Trees

-

JSBACH

-

- +

trees as fuel

+

trees as fuel

+

trees as fuel

+

water use by grass (only transitory)

damage to trees

(a)

0 10 20 30 40 50 60 70 80 90 100

0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000

mean annual rainfall (mm y⁻¹)

tree cover (%)

A B

(b)

JSBACH LPJ-Guess-Spitfire aDGVM

Extent Global Global Mainly tropical

Coupled/offline Coupled model Only off line Only off line

Type Vegetation Plant-Funct-Type Individual based Individual based

Length Run CMIP5-run 1850-2005 1960-2007 TRMM CRU

Resolution 1.9° 1° 1.0°

Grass-Fire-Feedback Negative Positive / Negative Positive

Results

Water limitation î Yes Yes Yes

Grass-Tree Competition Trees always outcompete grass At low P grass outcompete grass Grasses better competitors via seedlings

Bimodality No No No

î

0 20 40 60 80 100

0 500 1000 1500 2000 2500 3000

B

C

0 20 40 60 80 100

0 500 1000 1500 2000 2500 3000

B

C

î

0 20 40 60 80 100

0 500 1000 1500 2000 2500 3000

B

C

0 20 40 60 80 100

0 500 1000 1500 2000 2500 3000

B

C

0 20 40 60 80 100

0 500 1000 1500 2000 2500 3000

B

C

î

0 10 20 30 40 50 60 70 80 90 100

0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000

tree cover (%)

A B

Faculty of Geosciences Copernicus Institute of Sustainable Development Environmental Sciences