Supporting information for Three centuries of dual pressure from land use and climate change on the biosphere

Transcription

1 Supporting information for Three centuries of dual pressure from land use and climate change on the biosphere Sebastian Ostberg 1, Sibyll Schaphoff 1, Wolfgang Lucht 1,2 and Dieter Gerten 1 1 Research Domain 1 Earth System Analysis, Potsdam Institute for Climate Impact Research, Telegraphenberg A62, D Potsdam, Germany 2 Department of Geography, Humboldt-Universität zu Berlin, Berlin, Germany sebastian.ostberg@pik-potsdam.de S1. V metric The V metric measures the difference in vegetation structure in terms of the importance of broad life form types (grass, trees, bare ground), further characterized by their assigned attributes (Sykes et al., 1999). V (i, j) = 1 { [ min(v ik, V jk ) 1 ]} (ω kl a ikl a jkl ) (1) k l V ik and V jk describe the area fractions covered by life form k in ecosystem i and j, a ikl and a jkl are the attributes l of lifeform k in ecosystem i and j, respectively. Attributes are weighted for each life form by ω kl. Attributes can be climatic (tropical, temperate, boreal), or phenologic (evergreen, deciduous) or describe leaf types (needleleaved, broadleaved). A new attribute naturalness is added to distinguish between natural vegetation and croplands and pastures. Table S1 lists plant-functional types (PFT) and crop-functional types (CFT) simulated by LPJmL with their associated lifeforms and attributes. PFT abbreviations : TrBE, tropical broadleaf evergreen tree; TrBR, tropical broadlead raingreen tree, TeNE, temperate needleleaf evergreen tree; TeBE, temperate broadleaf evergreen tree; TeBS, temperate broadleaf summergreen tree; BoNE, boreal needleleaf evergreen tree; BoS, boreal summergreen tree. S2. Crop management in LPJmL Different forms of crop management are not eplicitly modelled in LPJmL. In order to calibrate simulated yields to match production statistics, e.g. by the United Nations Food and Agriculture Organization (FAO), LAI ma, the maimum leaf area inde of crops, can be scaled between 1 and 7, acting as a proy for planting density and fertilizer application. When calibrated, developed countries have mostly high LAI ma values, fast developing countries have middle LAI ma values and developing countries have low LAI ma values (Fader et al., 2010). We use a fied setting of LAI ma =5 for the whole period and all crops in all our simulations because historical management

2 Supporting information 2 intensities are not well-documented. Figure S1 tests the sensitivity of the Γ metric to different values of LAI ma. It shows that the of different management intensities on Γ is at least one order of magnitude smaller than the measured land use change. Differences are largest for the lowest LAI ma setting of 1, which increases the global mean land use change for the period from 0.11 to The area eposed to major LULCC impacts increases from 15.5 to 16.6%. Differences for the middle and high LAI ma settings are even smaller. Looking at the combined full impact of LULCC and CC, the uncertainty from management intensities is well within the range of uncertainty ca by the 20 climate realizations. S3. Decomposition into metric components The full Γ metric measures the magnitude of change, but provides little insight into the type of change taking place in response to climate or land use. Figure S3 shows how the four components ecosystem balance b S(b, σ b ), global importance g S (g, σ g ), local change c S (c, σ c ) and vegetation structure V S ( V, σ V ) from equation 1 in the main tet add up to Γ in each biome. To illustrate the relative contributions of different processes, we also provide the combination of ecosystem balance, local change and global importance computed for the variable subsets carbon stocks, carbon echange flues and water echange flues from table 1 in the main tet. We provide results for the full impact as well as the land use change and climate change. For eample, LULCC has a far larger impact on water flues than climate change in most biomes. This is because LULCC causes a strong shift from productive (transpiration) to unproductive water use (evaporation from bare soils). The shifts in vegetation structure ca by LULCC are also larger than those ca by climate change, with a few eceptions: there is a climate and CO 2 -driven epansion of tropical forests and woody savanna into some warm (open) savannas and grasslands, which causes strong compositional shifts. Infilling of sparse tree populations in the tundra also results in changes of vegetation structure at a magnitude that is otherwise only achieved by LULCC. S4. Biome classification scheme Biome classification in this study is based on a modified version of the scheme presented in Ostberg et al. (2013). It is based primarily on the composition of PFTs modelled in LPJmL. Compared to our earlier work, tree cover limits for the different savanna types (warm woody savanna & woodland, temperate woody savanna & woodland, warm savanna and temperate savanna) have been aligned with the IGBP classification scheme (Friedl et al., 2002). Also, Ostberg et al. (2013) an additional vegetation carbon limit to distinguish between tropical forests and warm woody savannas, which has been replaced by a tree leaf area inde (LAI) limit to achieve better results under pre-industrial conditions (figure S5).

3 Supporting information 3 Table S1. Plant-functional types and crop-functional types with their assigned attributes. Lifeform Attributes Tree: Evergreenness Needleleavedness Tropicalness Borealness Naturalness TrBE TrBR TeNE TeBE TeBS BoNE BoS (attribute weights: ) Grass: Tropicalness Naturalness C3 grass 0 1 C4 grass 1 1 Temperate Cereals 0 0 Rice 1 0 Maize 1 0 Tropical Cereals 1 0 Pulses Temperate Roots 0 0 Tropical Roots 1 0 Sunflower Soybean 1 0 Groundnut 1 0 Rapeseed Sugarcane 1 0 Others Managed grass 0 (attribute weights: ) BoS primarily represents broadleaved trees, but includes larchs. Derived from relative share of C4 grasses as determined dynamically by LPJmL

4 Supporting information 4 Figure S1. Impact of management settings in LPJmL on Γ. Default LAIma =5 for the analysis in this study. Difference in Γ values for LAIma settings of 1, 3 and 7 plotted for the end ( ) of the simulation period. Values in brackets denote range across 20 different climate realizations. Analysis of management impacts done for one of the 20 climate realizations.

5 Supporting information 5 change % 97.5% <=5 99.5% <=9 2% 0.5% change n: m: % 94.5% <=5 99.2% <=9 3.1% 1.2% 8 n: m: % 88.8% <=5 98.1% <=9 99.9% 5.1% 1.9% change n: m: % 76.9% <=5 94.9% <=9 99.9% 9.1% 3.8% Cumulative fraction of global ice change change n: m: % 63.4% <=5 81.5% <=9 99.5% 13.9% 8.6% change n: m: % 56.9% <=5 71.9% <=9 96% 15.7% 15.4% Fraction of landscape transformed Γ metric n: m:0 Figure S2. Historical epansion and intensification of land use. Left panels show land use change at the landscape scale (ensemble mean across 20 climate realizations) while right panels show cumulative land use density for the corresponding time frame.

6 Supporting information 6 Tropical Rainforest change Tropical Seasonal & Deciduous Forest change Pre industrial biome distribution Warm Woody Savanna & Woodland Temperate Broadleaved Evergreen Forest change Warm Savanna change Temperate Broadleaved Deciduous Forest change Warm Grassland change Mied Forest change change Desert Temperate Coniferous Forest change change Temperate Woody Savanna & Woodland change Temperate Savanna change Temperate Grassland change Boreal Evergreen Forest Boreal Deciduous Forest Tundra change change change carbon flues carbon stocks water flues ecosystem Γ metric balance global vegetation importance local structure change Figure S3. Decomposition of Γ values per biome. Components ecosystem balance b S(b, σ b ), global importance g S (g, σ g), local change c S (c, σ c) and vegetation structure V S ( V, σ V ) are combined into the full Γ metric. Values for carbon stocks, carbon flues and water flues illustrate the relative contribution of different processes to the full metric. Results shown for the period Colours in biome titles correspond to colours in the map.

7 Supporting information 7 a) b) fraction fraction ma. value: 1 ma. value: 1 land use fraction Figure S4. Total land use fraction. Maps show land use fraction (sum of croplands and managed grasslands/pastures) corresponding to impacts shown in figure 2 in the main tet.

8 Supporting information 8 Fractional coverage of PFTs Tree leaf area inde (LAI) Temperature Total tree cover >6 BoNE + BoS +TeNE 6 of total tree cover BoNE dominant Boreal evergreen forest Boreal deciduous forest BoS dominant Temperate coniferous forest Tropical trees < & BoNE + BoS + TeNe < of total tree cover TeBE share > TeBS share Temperate broadleaved evergreen forest Temperate broadleaved deciduous forest Tropical trees >6 of total tree cover Tree LAI > 3 TrBE > TrBR Tropical rainforest Tropical seasonal & deciduous forest Mied forest T annual -2 C Total tree cover 3 C4 grass share C3 grass share Warm woody Savanna & woodland Temperate woody Savanna & woodland Warm savanna Total tree cover 1 C4 grass share C3 grass share Temperate savanna Warm grassland Total vegetation cover >5% C4 grass share C3 grass share Temperate grassland Desert Tundra Figure S5. Biome classification scheme. For PFT abbreviations see section S1. Modified after Ostberg et al. (2013).

9 Supporting information 9 Tropical Rainforest ( million km²) Warm Woody Savanna & Woodland ( million km²) Temperate Broadleaved Evergreen Forest ( million km²) Temperate Woody Savanna & Woodland (3 3.3 million km²) Boreal Evergreen Forest Tropical Seasonal & Deciduous Forest ( million km²) Warm Savanna ( million km²) Temperate Broadleaved Deciduous Forest (0.9 1 million km²) Temperate Savanna (3 3.1 million km²) Boreal Deciduous Forest Pre industrial biome distribution Warm Grassland ( million km²) Mied Forest ( million km²) Temperate Grassland ( million km²) Tundra Desert ( million km²) Temperate Coniferous Forest ( million km²) ( million km²) ( million km²) ( million km²) Severity on separate areas major impacts on areas of land use or natural vegetation moderate impacts on areas of land use or natural vegetation Figure S6. Human transformation of natural ecosystems across biomes. Impacts on land use areas shown from above. Analogous to figure 1a) in the main tet these are calculated just for the cultivated fraction, not at the landscape level. Impacts of climate change on remaining natural ecosystems shown from below.

10 REFERENCES 10 Figure S7. Simulated tree cover in the high latitudes. Black line denotes tree line according to Brown et al. (1998). References Brown, J., Ferrians Jr., O.J., Heginbottom, J.A. and Melnikov, E.S. (1998). Circum-arctic map of permafrost and ground-ice conditions, Bolder, Co: National Snow and Ice Data Center. Digital media. data/ggd318 Fader, M., Rost, S., Müller, C., Bondeau, A. and Gerten, D. (2010). Virtual water content of temperate cereals and maize: Present and potential future patterns, J. Hydrol. 384(3-4): /j.jhydrol Friedl, M., McIver, D., Hodges, J., Zhang, X., Muchoney, D., Strahler, A., Woodcock, C., Gopal, S., Schneider, A., Cooper, A., Baccini, A., Gao, F. and Schaaf, C. (2002). Global land cover mapping from MODIS: algorithms and early results, Remote Sens. Environ. 83(1-2): Ostberg, S., Lucht, W., Schaphoff, S. and Gerten, D. (2013). Critical impacts of global warming on land ecosystems, Earth Syst. Dyn. 4(2): Sykes, M. T., Prentice, I. C. and Laarif, F. (1999). Quantifying the impact of global climate change on potential natural vegetation, Clim. Change 41(1):