17/01/2006 1
J.C. Gégout, C. Piedallu, I. Seynave, J.C. HervéAgroParisTech - INRA - IFN
Large scale mapping of soil pH by plant presence/absence
bioindication
49th Annual Conference of the IAVS
2
Plant distribution models are built with climatic variables and sometimes with proxies of soil resources (soil types, geology)
Nutritional variables are of major importance to plant growth and distribution
Measures of soil variables are expensive. Thus, it is difficult to gather enough measurements to make accurate maps
Is plant species bioindication useful to map nutritional soil resources ?
The need of nutritional direct variables for plant distribution modelling
3
Data
French National Forest Inventory (IFN)Systematic sampling in forests88 004 floristic relevés without pH measures
EcoPlant3 835 floristic relevés with pH measures
4
Response curve
Ecological variable
1
0
Pres/abs
Ecological variable
1
0
Response curve
Indicator value
IV
EcoPlant Indicator values
IV for 568 frequent plant species
5
Predicted value : mean of species IVEllenberg (1974), ter Braak & Barendregt (1986)
5.7 3.0
6.2
8.3
6.5
Predicted pH : 6.0
0
0.1
0.2
0.3
pH4.0 4.9 5.9 6.9 7.9
Dryopteris filix-mas
Melica uniflora
Sambucus nigra
Bioindication with IV
6
Mean of IV on each plot
88 004 IFN plots + pH_IV
88 004 IFN plots
with floristic inventory
pH prediction on IFN plots
7
3.0 - 3.53.5 - 4.04.0 - 4.54.5 - 5.05.0 - 5.55.5 - 6.06.0 - 6.56.5 - 7.07.0 - 7.57.5 - 8.08.0 - 8.5lack of data
pH classes
0 500 kms
IDW Interpolation
pH mapping
8
Map validation
261 validation plots on a grid of 16 x 16 kmsRepresentative of French Forest
R2 = 0.56, REQM = 0.82
3 4 5 6 7 8Mapped pH
3
4
5
6
7
8
Mea
sure
d pH
Measure of prediction errorsR2
REQM = (1/N * (X - X2)^
9
Plots variables Sampling Plots nbStep
PrincipleA few expensive relevés with measures of variables to establish plant IVA lot of cheap relevés without measures to map variables
Plant species & measured variables
Stratified according to x
fewIndicator value
Plant species Systematic largeMapping
Variables mapping by bioindication
Systematic fewMap quality Plant species & measured variables
10
The databases « IV », « mapping » and « quality » are constructed Acidity (pH)
Nitrogen avail.(C/N)
It is now easy to build up new maps of resources with other EcoPlant IV
IFN establish 7 000 - 10 000 new plots/year-> Increase of map resolution-> Possibility of monitoring by comparing maps of different periods
A generalisable method
Towards new distribution models integrating both climate and soil direct variables
17/01/2006 11
Thank you !
12
Comparison of pH and C/N Maps
13
PrésenceAbsence
Validation data (88 004 plots)
Map of the potential distribution of Acer campestre
Nutritional and climatic model
Climatic model
Climatic model, success: 54 %Nutritional and climatic model, success: 73 %
14
Map of the Beech potential productivity