ENGREF: French Institute of Forestry, Agricultural and Environmental Engineering INRA: French National Institute for Agricultural Research
48th IAVS Symposium, Ecoinformatics special session Lisbon, Portugal - 25 July 2005
Predicting the distribution of Acer campestre (L.) with climatic and nutritional factors in France Christophe COUDUN*, Jean-Claude GEGOUT and Christian PIEDALLU *Present
address: Centre for Terrestrial Carbon Dynamics, CTCD, UK,
[email protected]
Should we incorporate nutritional factors in complement to climate in species distribution models? Many (hundreds of…) recent studies on species distribution modelling with environmental predictors (Scott et al. 2002, Guisan et al. 2002, Lehmann et al. 2002) Various purposes: invasion issues, consequences of global change, biogeographic hypotheses, conservation issues...
Most studies (95 %…) only use climatic explanatory factors and do not explore the contribution of soil nutritional variables Development of EcoPlant, database for French forests, to take simultaneously climatic and edaphic factors into account
Choice of modelled species: Field Maple (Acer campestre L.) Species which can be easily recognised in the field, relatively unmanaged by foresters, and which reacts to soil conditions 25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
2
EcoPlant: phytoecological database for French forests
120 data sources 6 500 relevés 2 000 species (700 are frequent)
200 climatic and edaphic variables (pH, Ca, Mg, K, Al, H)
Gégout, J.-C., Ch. Coudun, G. Bailly and B. Jabiol (2005): EcoPlant: a forest site database linking floristic data with soil and climate variables. Journal of Vegetation Science, 16(2), 257-260.
25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
3
Calibration data set (EcoPlant)
3,286 relevés 460 relevés with presence of Acer campestre (black points) 156 climatic variables (T, P, radiation, PET, AET, WB…) 6 edaphic variables (pH, C/N, S/T, Ca, Mg, K) 25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
4
Evaluation data set (IFN)
88,004 relevés from the French National Forest Inventory 14,373 relevés with presence of Acer campestre (green points)
presence absence
25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
5
Ecological response of Acer campestre (forward stepwise logistic regression)
I. Climatic model only
II. Climatic and edaphic model
1. Autumn precipitation (pAu) 2. Annual actual evapotranspiration (AETTh)
1. soil pH (pH) 2. Autumn precipitation (pAu) 3. Annual actual evapotranspiration (AETTh)
prob.(acer) = f(pAu, AETTh)
prob.(acer) = f(pH)
Preference for rich pH
Preference for high AETTh Preference for low pAu
pH
25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
6
Climatic maps for model visualisation (e.g., mean temperature) (AURELHY meteorological model)
Maps for pAu and AETTh also available 25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
7
Construction of a map of soil pH in French forests 1. Computation of pH optima (indicator values) for 700 frequent species
2. Estimation of soil pH on 88,004 IFN sites based on species indicator values e.g., north-east France
(Gegout, Coudun, Brisse and Berges 2002)
vamy : Vaccinium myrtillus anne : Anemone nemorosa coma : Convallaria majalis
low pH high pH
optimum : value of the gradient where p is maximal 25 July 2005
3. Spatial interpolation of bio-indicated soil pH values
Predicting the distribution of Acer campestre (L.) in French forests
8
Resulting map of soil pH (spatial interpolation of bio-indicated values, from EcoPlant)
25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
9
Map 1. Predicting the distribution of Acer campestre with climatic factors only (pAu + AETTh)
presence absence
Success = 54 % 73 % presence, 50 % absence
25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
10
Map 2. Predicting the distribution of Acer campestre with climatic and edaphic factors (pH + pAu + AETTh)
presence absence
Success = 73 % 86 % presence, 70 % absence
25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
11
Modelling plant species distribution: the importance of soil factors Climate plays an important role at large geographical extents Climate allows to predict species geographic range
Soil acts as a local filter at smaller geographical extents Soil improves the prediction of species presence/absence inside its climatic geographic range Coudun, Ch. and J.-C. Gégout (submitted): Nutritional factors are needed in complement to climate in predicting plant species distribution: an illustration with Acer campestre L. in France. Journal of Biogeography. 25 July 2005
Predicting the distribution of Acer campestre (L.) in French forests
12