SAFE Fourth Year Annual Report Volume 3

Quality of Life and Management of Living Resources
Silvoarable Agroforestry For Europe (SAFE)

European Research contract QLK5-CT-2001-00560
Final progress report covering the period August 2004 to 31 January 2005
1.4
Poplar Cherry
1.2 Oak Pine
Walnut
Relative tree yield
1.0
0.8
0.6 7
Tree transpiration
0.4
6 Crop transpiration
Soil evaporation
0.2
5
0.0
mm day-1
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 4
Relative crop yield

3
0
01/01/2002 01/04/2002 30/06/2002 28/09/2002 27/12/2002 27/03/2003 25/06/2003 23/09/2003
1.0
Relative crop yield
0.8
0.6
0.4
0.2
0.0
0 10 20 30
Arable 156 trees/ha

Silsoe Leeds
SAFE PROJECT FINAL PROGRESS REPORT

1
Volume 3 : Contractor Reports, May 2005
Foreword
This volume includes the contributions of the 10 Contractors of the SAFE project during the
fourth year of the project. Most of the information is included here. For a rapid survey of the
achievements of the SAFE consortium, please refer to Volume 1 (Synthesis) and Volume 2
(Work-Package reports).
Service tree (Sorbus domestica L.) is a promising agroforestry tree that was however not
considered during the SAFE project.
2
TABLE OF CONTENTS
1 CONTRACTOR 1: INRA – SYSTEM ........................................................................................ 6
SCIENTIFIC TEAM AND TIME SPENT ON THE WPS ................................................................................ 6
CONTRIBUTION TO WORKPACKAGES ................................................................................................... 8
WP2: European Silvoarable knowledge.......................................................................................... 8
WP3 Silvoarable experimental network .......................................................................................... 9
WP4 Above-ground interactions ................................................................................................... 21
WP5 Below-ground interactions ................................................................................................... 22
WP6a Biophysical integrated plot model Hi-sAFe ....................................................................... 38
WP6b: Simple biophysical model Yield-sAFe ............................................................................... 44
WP7. Economics of silvoarable agroforestry................................................................................ 56
WP10. Project co-ordination ........................................................................................................ 56
2 CONTRACTOR 1: INRA-AMAP ............................................................................................. 90
SCIENTIFIC TEAM ............................................................................................................................... 90
TIME SPENT ON THE DIFFERENT WORKPACKAGES............................................................................. 91
CONTRIBUTION TO WORKPACKAGES ................................................................................................. 91
WP4. Above-ground interactions (0.6 person-months) ................................................................. 91
WP6. Modelling (5.6 person-months) ........................................................................................... 91
DISSEMINATION ................................................................................................................................. 92
Participation in meetings and workshops ..................................................................................... 92
Scientific publications ................................................................................................................... 92
Utilisation de maquettes architecturales de Noyers hybrides pour Hi-sAFe. (in French)............ 92
References ................................................................................................................................... 105
3 CONTRACTOR 1: INRA – UAFP .......................................................................................... 106
SCIENTIFIC TEAM AND TIME SPENT ON THE WPS ............................................................................ 106
4 SUB-CONTRACTOR TO INRA :ICRAF .............................................................................. 130
5 SUB-CONTRACTOR TO INRA :CTL................................................................................... 132
1 . Installation de sites expérimentaux :...................................................................................... 133
2 . Suivis et mesure de sites expérimentaux : .............................................................................. 133
6 CONTRACTOR 2: WAGENINGEN UNIVERSITY ............................................................ 135
CONTRIBUTION TO WORKPACKAGES ............................................................................................... 135
SUB-CONTRACTOR FINIS E.V.: CONTRIBUTIONS TO WP9.............................................................. 136
SUB-CONTRACTOR GPG: CONTRIBUTION TO WP9......................................................................... 137
7 CONTRACTOR 3: NERC - CENTRE FOR ECOLOGY AND HYDROLOGY................ 141
WP5 Below-ground interactions ................................................................................................. 141
SAFE Final Progress Report – Volume 3 – May 2005 3

WP8 Scaling up to the farm and region ...................................................................................... 142
WP9 European guidelines for agroforestry ................................................................................ 142
Significant difficulties during the reporting period..................................................................... 147
8 CONTRACTOR 4: UNIVERSITY OF LEEDS ..................................................................... 148
2. SCIENTIFIC TEAM AND TIME SPENT ON THE WPS ........................................................................ 148
1. Partner number, name and address of the participating organisation ................................... 148
3. Time spent on the different workpackages during year 4 (months). ....................................... 148
4. CONTRIBUTION TO WORKPACKAGES ........................................................................................... 149
4.1 WP1 Silvoarable modelling strategies .................................................................................. 149
4.2 WP2 European silvoarable knowledge ................................................................................. 149
4.3 WP3 Silvoarable experimental network ................................................................................ 149
4.3.1. T3.1: Collect data from existing experiments as required by the modelling activity ........ 150
4.3.2 At the SAFE experimental sites specific information needed to parameterise the biophysical
model will be collected ................................................................................................................ 153
4.3.3. Results and deliverables .................................................................................................... 153
4.3.4. Future work ....................................................................................................................... 153
5. OTHER WORK ............................................................................................................................... 154
6. SIGNIFICANT DIFFICULTIES OR DELAYS EXPERIENCED DURING THE REPORTING PERIOD ........... 154
7. DISSEMINATION ........................................................................................................................... 154
9 CONTRACTOR 5: CRANFIELD UNIVERSITY ................................................................. 155
SCIENTIFIC TEAM ............................................................................................................................. 155
TIME SPENT ON THE DIFFERENT WORK-PACKAGES ......................................................................... 155
CONTRIBUTION TO WORK-PACKAGES ............................................................................................. 155
WP6B. Minimal biophysical integrated model: Yield-sAFe........................................................ 155
WP7. Economic modelling at a plot-scale .................................................................................. 158
WP8. Scaling up to the farm and region .................................................................................... 158
References ................................................................................................................................... 159
10 SUB-CONTRACTOR TO CRAN : BEAM ........................................................................ 160
11 CONTRACTOR 6: CNR-PORANO.................................................................................... 161
2. SCIENTIFIC TEAM ......................................................................................................................... 161
2.1. Principal Investigators ......................................................................................................... 161
3.TIME SPENT ON THE DIFFERENT WORKPACKAGES........................................................................ 162
4. CONTRIBUTION TO WORKPACKAGES:.......................................................................................... 162
4.1. WP2: European Silvoarable Knowledge.............................................................................. 162
4.2. WP3. Silvoarable experimental network .............................................................................. 163
4.3. WP4 Above ground interactions........................................................................................... 170
4.4. WP9 EU guidelines .............................................................................................................. 172
12 CONTRACTOR 7: UNIVERSITY OF EXTRAMADURA .............................................. 177
PERSON-MONTHS..................................................................................................................... 178
WP2. EUROPEAN SILVOARABLE KNOWLEDGE ................................................................................ 178
WP3. SILVOARABLE EXPERIMENT WORK........................................................................................ 178
WP4. Above-ground interactions ................................................................................................ 179
WP5. Below-ground interactions ................................................................................................ 179
WP8. Scaling-up to the farm and the region ............................................................................... 180
WP9. European guidelines for policy.......................................................................................... 180
SUB-CONTRACTED WORK TO FGN .................................................................................................. 181
EXPLOITATION AND DISSEMINATION ACTIVITIES ............................................................................ 181
LIST OF ANNEXES............................................................................................................................. 181

13 CONTRACTOR 8: FAL....................................................................................................... 183
CONTRIBUTION TO WORKPACKAGES .............................................................................................. 183
WP1. Silvoarable modelling strategies (2.0 person-months) ...................................................... 183
WP2. European silvoarable knowledge (2.0 person-months) ..................................................... 183
WP6. Biophysical integrated plot modelling (1.0 person-month) ............................................... 183
WP7. Economic modelling at the plot scale (2.0 person-month) ................................................ 184
WP8. Scaling-up to the farm and the region (48.0 person-months)............................................ 184
WP9. European guidelines for policy implementation (5.3 person-months) .............................. 184
FUTURE WORK ................................................................................................................................. 185
SIGNIFICANT DIFFICULTIES OR DELAYS EXPERIENCED DURING THE REPORTING PERIOD ............... 185
14 CONTRACTOR 9: APCA.................................................................................................... 186
WP2 European silvoarable knowledge (10 person-month)......................................................... 186
WP7 Economic modelling at the plot scale (9.0 person-months)................................................ 189
WP8 Scaling-up to the farm and the region (14.0 person-months)............................................. 203
WP9 European guidelines for policy implementation (8.4 person-months) ............................... 204
15 CONTRACTOR 10: UNIVERSITY OF THESSALONIKI.............................................. 215
SCIENTIFIC TEAM AND TIME SPENT OF THE WPS............................................................................. 215
Workpackage 2: European silvoarable knowledge ..................................................................... 215

INRA Report
1 Contractor 1: INRA – SYSTEM

Name and address of the participating organisation
Contractor 1: INRA – SYSTEM (FRANCE)

INRA-SYSTEM, Systèmes de Culture Méditerranéens et Tropicaux,
2 Pl. Viala, 34060 Montpellier Cedex 1, France
Scientific team and time spent on the WPs

Principal investigators
Name. Unit Tel Fax E-mail

Dr. Dupraz Christian SYSTEM 33 4 99 61 23 39 33 4 67 52 21 16 dupraz@ensam.inra.fr
Lecomte Isabelle SYSTEM 33 4 99 61 25 52 33 4 67 52 21 16 lecomte@ensam.inra.fr
Dr. Dufour Lydie SYSTEM 33 4 67 61 75 71 33 4 67 61 55 12 lydie.dufour@cirad.fr
Mineau Jonathan CTL 33 4 99 61 25 52 33 4 67 52 21 16 mineau@ensam.inra.fr
Mulia Rachmat SYSTEM 33 4 99 61 26 84 33 4 67 52 21 16 mulia@ensam.inra.fr
Time spent on the different work packages during the final 6 months
For the purpose of the project coordination, this table includes some time for Christian
Dupraz and Isabelle Lecomte for the project coordination in February and March 2005, as
allowed by the technical annex of the contract and approved by the project officer in
Brussels.
INRA Staff
Name Unit WP1 WP2 WP3 WP4 WP5 WP6 WP7 WP8 WP9 WP10 Total
Dupraz Christian SYSTEM 0,5 1 3,1 1,75 0,2 3,5 10,05
Lecomte Isabelle SYSTEM 1 4 6 1 12
Mulia Rachmat SYSTEM 2 10 12
Dufour Lydie SYSTEM 2 1 1 4
Martina Mayus SYSTEM 0,7 1,5 0,2 0,6 5,5 0,2 0,3 9
Total INRA-SYSTEM 0 0,7 6 3,2 18,7 13,25 0,2 0 0,5 4,5 47,05
INRA-SYSTEM sub-contractors
Jonathan Mineau CTL 4 4
Jean-Philippe Terreaux CHAV 6 6
Meine van Noordwijk ICRAF 0,5 0,5
Grégoire Vincent ICRAF 2 2 4
Betha Lusiana ICRAF 1,2 1,2
Degi Harga ICRAF 1,5 1,5
Total 0 0 4 2 3,2 2 6 0 0 0 17,2
Grand total 0 0,7 10 5,2 21,9 15,25 6,2 0 0,5 4,5 64,25
The final total of INRA-SYSTEM time allocation to the whole SAFE project is 120.5 person-
month of INRA staff (compared to 105 indicated in the contract) and 43 person-months of
sub-contractor staff (identical to the contract).
SAFE Final Progress Report – Volume 3 – May 2005

Wild cherry trees responded spectacularly to the
The Swiss team leader Felix Herzog gifted the
localized nitrogen experiment at INRA-SYSTEM.
project coordinator Christian Dupraz with this
Nitrogen was localized in only two voxels resulting in a
superb book on extra-ordinary trees of the world
parallelepiped rooting pattern. Walnut trees did not react
at the Zurich final workshop of the SAFE project
as strongly.
A view of the split-root experiment with hybrid
walnut trees that allowed to calibrate the The SAFE project 2005 greeting card was prepared by
responsiveness of the tree root system to an INRA-SYSTEM as part of WP10 (coordination) activity
heterogeneous environment
A view of the chair of the SAFE final national Conference
Measurements of poplar sap flow at the at Paris on January 26 2005.
Vézénobres experimental plot in collaboration with From left to right : Michel Delacroix, President of the
colleagues from CNRS in Toulouse (Etienne forest commission of APCA, Luc Guyau, President of
Muller and Luc Lambs) APCA, Christian Dupraz, coordinator of the SAFE project
and Fabien Liagre, SAFE WP2 leader

Contribution to work packages
WP2: European Silvoarable knowledge
While being contracted by INRA-System, Martina Mayus kept monitoring the Netherlands
sub-contractors of WU. Here are some details of this activity during the last period of the
contract.
Supervision of the literature study of the MSc student Michel Postma: “Its’a all in the mix. -
Agroforestry, a prospective land use system for the Netherlands”. The overall aim of this
report is to explore the potential of agroforestry as an innovative and sustainable multiple
land use system in the Netherlands. Agroforestry systems with prospects for the
Netherlands, appropriate tree and crop species as well as required regulations and subsidies
with respect to Agroforestry were identified. Four distinct agroforestry practices (Figure 2 and
3) are considered to be appropriate and innovative for The Netherlands, namely:
1. silvoarable agroforestry, comprising widely-spaced trees and/or shrubs associated

with arable crops
2. silvopastoral agroforestry, a combination of trees, forage (pasture) and livestock
3. forest gardening, comprising multi-species and multi-storied dense plant associations,

planted and/or managed in such way that they mimic the structure and the ecological
processes of natural forests
4. forest farming, the cultivation of edible, medicinal or decorative specialty crops as

under-storey in (semi)natural woodlands.
Table 1. Windbreaks and riparian buffers are common practices in the Netherlands.

Table 2. Innovative silvoarable (left) and silvopastoral (right) agroforestry systems
appropriative in particular for Dutch farmers.
Table 3. Innovative agroforestry systems, i.e. forest farming (left) and forest gardening
(right), appropriative in particular for Dutch foresters and hobby farmers, respectively.
The literature study and the results of the earlier performed survey on farmers’ attitude
towards agroforestry were synthesised by setting up agroforestry design scenarios for
farmers, foresters and hobby farmers. For each type of land user several examples of
prospective agroforestry designs and practices are suggested.
It was concluded that agroforestry fits well into the current Dutch governmental policies
aiming for extensification of agricultural land, leaving more room for nature and thus
increasing biodiversity. However, two major constraints exists: 1) Agroforestry is not a
recognised land status and no subsidies are currently available. 2) There is a lack of
adequate research, demonstrations and information towards land users and policy makers.
Most farmers and policy makers are not aware of the possibilities of agroforestry. The report
was presented in a seminar at the WU department Plant Production, Wageningen (Safe
website).
WP3 Silvoarable experimental network

The field activities of INRA-SYSTEM are concentrated on two different silvoarable systems
not far from Montpellier, Southern France. At Restinclières walnut stands and at Vézénobres
poplar plantations were combined with durum wheat during the last two SAFE-project years.
The growing season of 2003 revealed two interesting results: First, the combination of walnut
and the winter crop durum wheat is a good choice in terms of limiting competition. The
Hybrid walnut tree is characterised by a late bud break in spring. By the time that the walnut

trees are full in leave, durum wheat is already far in its development. Overall, the influence of
walnut on the winter crop was negligible. Secondly, at Vézénobres, different from what was
expected for a Mediterranean region, the competition for light was playing a more dominant
role than water competition. In 2004, the experiments were designed to explore the research
findings of the prior season. In general the studies at Vézénobres were intensified compared
to Restinclières. At both sides, experimental protocols and management actions were
performed as similar as possible.
Site Tree Year of Crop in 2003- Compared AF treatments

plantation 2004 treatments (number of modalities)
Restinclières Hybrid 1995 Durum wheat Forest control Tree-crop distance (2)
walnut (Claudio variety) Crop control Tree line orientation (2)
Agroforestry
Vézénobres Poplars 1996 and Durum wheat Forest control Tree-crop distance (2)
1997 (Durango Crop control Tree-line orientation (2)
variety) Agroforestry Tree canopy pruning (3)
Table 4: Main features of the two experimental plots managed by INRA-SYSTEM
Climate and environmental site characterisation at both INRA sites
Climate and microclimate recording

At both sites, INRA-SYSTEM meteorological stations record hourly data of air temperature,
air humidity, photosynthetically active radiation and rainfall. Both stations are set up at a
minimum distance of 30 m from the trees in the experimental agroforestry plots, in order to
record the boundary climate outside the influence of the trees. For the following years, it is
considered to move the stations further from the trees, as they grow higher.
Soil humidity and water table

The soil water content was measured with neutron probes during as well as outside the
growing season at two weekly intervals. The water table level is needed to compute a correct
water budget of the silvoarable system with the Hi-sAFe model. INRA-SYSTEM, therefore,
equipped the plots at Restinclières (in 2002) and Vézénobres (in 2003) with piezometers.
Soil fertility
The arable crop has been well fertilised with nitrogen (about 180 kg N ha-1 year –1, ca. 450 kg
ammonium nitrate/ha) during all years. Therefore, nitrogen should have been sufficient and
competition for nitrogen should have been negligible. However, if cropping operations are not
homogeneous within the agroforestry plot, tree impact on crop yield could be wrongly
evaluated. We, therefore, measured the homogeneity of the nitrogen fertiliser application at
both sites. Nitrogen fertiliser is spread with “rotatic” sprayers that usually need to be driven at
precise distances to allow cross-fertilisation at the margin. In agroforestry, this is not
possible, due to the fixed distance between tree rows. For the growing season 2002 -2003,
the distribution of the nitrogen application was measured in Restinclières and at Vézénobres.
We placed containers along transects and weighed the ammonium nitrate grains collected at
the different locations. Each transect consisted of 5 containers, i.e. one placed on the tree
row, and on each side of the tree row at distance of 2 m and in the centre of the alley (6.5 m
and 8 m from the tree rows at Restinclières and Vézénobres, respectively). In Restinclières,
for the growing season 2003 -2004, the first application of the nitrogen fertiliser was assumed
to be homogeneous, since performed with a pendular sprayer. The others were performed
with the usual machine. At Vézénobres the farmer used always a rotatic sprayer.
Tree size and phenological development

The impact of the trees on the intercrop depends on tree density, tree height and canopy
size, tree leaf area density and on the overlapping growth period of trees and crops. The

tree-crop interactions depend also on the degree of spatial and temporal complementarity. In
this context, the trees have been intensively measured in particular at Vézénobres:
phenology, height, diameter, leaf area, sap flow, and root length densities at different depths
and distances. The above- and belowground studies are described in detail in WP4 and
WP5, respectively.
200 walnut.ha-1 mixed with durum wheat 140 poplars.ha-1 mixed with durum
Walnut height ca. 7 m wheat
Poplar height ca. 20 m
Restinclières plot Vézénobres plot
2 tree row orientations x 2 sides x 2 distances 2 tree-row orientations x 2 canopy pruning
= 8 agroforestry treatments x 2 sides x 2 distances
2 replications in agroforestry A2 = 16 agroforestry treatments
2 replications in agroforestry A3 3 replications in agroforestry 1996
5 replications in crop control A2 3 replications in agroforestry 1997
3 replications in crop control of each plot
Table 5: The experimental agroforestry plots at Restinclières and Vézénobres and the
treatments of the season 2003-2004
The crop measurements

One objective of the crop measurements is to measure state variables of the wheat crop to
calibrate the STICS crop model used in Hi-sAFe. The model is an important tool to fully
integrate our knowledge and understanding of silvoarable systems and thus add to the
insights obtained from experiments (Report part WP6a). Secondly the field observations aim
to assess the influence of the trees on the growth and yield of durum wheat. This was
achieved by measuring its development and grain yields along transects of the tree-crop
interface and in the crop control, i.e. outside the influence of trees (Table 2). The influence of
trees on yield components was also determined. All measures were done at 2 distances and
2 orientations from the tree lines as given in Table 2. The experimental unit is a subplot
(micro-plot) of 1 m2, consisting of 7 to 8 one m long crop rows parallel to the tree line.

Site Distances Orientation Additional
from the tree crop treatments included
line
Restinclières 2 m and 6.5 m N-S (plot A2)
E-W (plot A3)
Vézénobres 2 m and 6.5 m N-S (plot 1997) Tree canopy pruning:
E-W (plot 1996) With and without
Table 6: Locations of crop growth measurements in agroforestry plots at Vézénobres
and Restinclières experimental plots
State variable Day of measure Day of measure Protocol
at Restinclières at Vézénobres
Tiller density n.a. 6.5.04 Number of tillers on the half the area of
2
the micro-plot (0.5 m ).
Phenological Every 5-10 days Every 5-10 days 5 randomly chosen plants per micro-plot
stage and plant (Zadoks scale) on 2 and 3 replications
height DOY 133 - 180 DOY 120 - 200. for Restinclières and Vézénobres.
22.5.04
Leaf area of n.a. and Length and 3 width of leaf 1 (flag leaf)
top leaves 17.6.04 and leaf 2.
Ear density 2.7.04 19.7.04 Ear number of all plants of the micro-
plot, counted at harvest.
Weed 10.3.04 Every 14 days Visual assessment in each micro-plot.
infestation 30.4.04
1.7.04
Larger weeding
Harvest 2.7.04 19.7.04 The micro-plots were harvested by
hand. Sub-samples were taken for grain
Wheat yield yield determination.
Weight of 1000 July 2004 July 2004 At INRA lab
grains
Protein content Not yet Not yet
Table 7: Calendar of crop measurements in agroforestry plots at Vézénobres and

Restinclières experimental plots
In Restinclières, only the intensive agroforestry association was investigated and at
Vézénobres, the crop yield measurements were focused on plots with trees of the poplar
clone I-214, who had been not root-pruned in 2002. We managed 2 and 5 replications in
agroforestry and control treatments in Restinclières, and at Vézénobres we had 3
replications for each treatment (Figure 1). Table 3 shows the timing of the various crop
measurements. The results of the walnut-wheat field studies in Restinclières have been
presented in the third Annual report of SAFE and will be compared, here, with the results of
the poplar wheat system at Vézénobres.

Durum wheat yields in the poplar agroforestry experiment at Vézénobres in
2004
Tree canopy pru ned Tree m icro-p lot
T ree row
Tree can opy u npruned
M icro-p lot
Table 8: Field layout with measurement points (micro-plots) at Vézénobres

The agroforestry system in Vézénobres consist of two silvoarable poplar stands, set up in
1996 and 1997 with tree rows in the North-South and East-West direction, respectively.
These plots are the most mature silvoarable sites in France or, even, in Europe. The poplar
plantations showed a fast growth in height and diameter, and we expect that its life cycle will
be not more than 10 - 12 years. This is short compared to the life cycles of the silvoarable
poplar plantations in Leeds and Cranfield (both in UK), which are expected to be at least 25
years.
During the first, i.e. the establishment years of the silvoarable field till season 2002, the
rotation of the intercrop differed between the two plots. In plot 1996, there was a rotation of
asparagus, durum wheat, durum wheat, sorghum, fallow, durum wheat; while in plot 1997
asparagus was planted during the first 5 years. Then in year 2002, both plots were left fallow
and during the last two growing seasons durum wheat was planted between the trees. The
effect of poplar on wheat was intensively investigated (wheat development, grain yield and
quality) at the different agroforestry treatments and compared with the measurements of the
monocrop control plots, one at the south edge of the 1996 plantation, and one at the west
edge of the 1997 plantation. The studies on durum wheat will continue during the growing
season of 2005.
Material and Method

Plot Area Treatments (area) Tree Tree row Wheat Poplar
(ha) spacing (m) direction variety clones
2003-2004 studied
1996 0.4116 Forest control 7x7 North-South no crop I-214
0.5400 Crop control no trees - Durango no trees
1.1520 Agroforestry 16 x 4.5 North-South Durango I-214
1997 0.5280 Forest control 8x6 East-West no crop I-214
0.2000 Crop control no trees - Durango no trees
1.4020 Agroforestry 15.5 x 4.5 East-West Durango I-214
Total 4.2336
Table 9: The Vézénobres site with 2 poplars plantation plots planted in 1996 and 1997
Climate and microclimate characterisation

At Vézénobres, the meteorological station was launched on day 155 in year 2003. The
station records hourly data of air temperature, air humidity, photosynthetical active radiation
(PAR) and rainfall. Wind velocity and global radiation are available at the Alès airport station,
located at less than 5 km from the experimental site. During both seasons, the station was
out of order for several days and sometimes for weeks. For these periods, the data from a
Météo-France weather station, at about 5 km from the experimental plots was used (Alès
Deaux airport).
The soil water volume (neutron tubes) as well as the water table (piezometers) was recorded
once a month between May and October 2004, with additional measurements after heavy
rainfalls all over the year. The neutron tubes and the piezometers were installed at the end of
November 2003 and end of April 2004 (DOY 330 and DOY 118), respectively. More detailed
information can be found in the report part of WP5.
Hemispherical photographs were taken to estimate the reduction of available light at a given
point in the intercrop, i.e. adjacent to the micro-plots at 2 and 6 m from the tree line. In 2004,
the photos were taken in May and June. The results of the prior season illustrate that the
available daily light is homogenous on the plot with a North-South orientated tree row and
heterogeneous for the plot with a West- East tree row orientation. The data will be used also
to validate the model Hi-sAFe. This requires information of the available radiation around an
average tree surrounded by average trees (torus symmetry), as can be obtained by
hemispherical photographs. The method was explained in the Second Year Contractor report
in the chapter of WP4.
Growth conditions
The agroforestry plot was well fertilised, so that nitrogen stress should be ignorable.
Furthermore, diseases, pests and weeds were controlled chemically, and the micro-plots
have been hand weeded when needed and just before the harvest (2004) (Table 3). The
poplars have access to water table, but this is not meaning that they are not suffering from
water stress.
Tree characteristics
In 2002-2003, a root pruning (= root trenching) treatment was set up in both plantations.
Roots were cut at 2 and 3.5 m from the tree line. This treatment appeared to have no impact
on the water competition between trees and crops and a limited effect on tree performance.
Therefore, the root pruning treatment was not further investigated during the growing season
2003 – 2004. The forestry control plots are disked twice a year to limit weed proliferation.
In April 2004, a new treatment was included in the agroforestry systems, i.e. two different
tree canopy pruning height (6 m is the standard, 10 m is the extra high pruning).

The poplar trees of both plots were measured for height and diameter at breast height (i.e. at
1.3 m) in January 2005. In November 2003, the poplars reached a height of about 20 m.
The major phenological dates, i.e. start, finish and date of 50% of a) bud break and b) leaf-
fall in were observed. In 2003, bud break started at the beginning of April (around DOY 95)
and leaf fall was monitored at the end of November (around DOY 327). In 2004, the dates
were very similar.
Tree management
Day of year Date Operation
Forestry plantation disked
April Extra high canopy pruning in blocks
Table 10: Calendar of tree management at the Vézénobres farm in 2003-2004
Durum wheat management

Day of year Date Operation
329 25.11.03 Sowing (150 kg ha-1)
5 1st week Jan Fertilisation (ammo nitrate: 180 kg ha-1)
56 Last week Feb Weeding (Hussard: 1l ha-1)
62 1st week March Fertilisation (sulfamo: 250 kg ha-1)
108 15 - 20 April Fertilisation (ammonitrate: 180 kg ha-1)
111 21 April Fungicide application (OPONAN)
120 / 182 30.4 + 1.7 Hand weeding of the micro plots
200 19.07.04 Harvest
Table 11: Calendar of durum wheat management at the Vézénobres farm in 2003-2004
Crop measurements
During the season 2003-2004, we concentrated the observations on plots with the poplar
cultivar I214. The root-pruning treatment was not repeated, since it showed no effect on
wheat production. A new treatment was included in the agroforestry systems, i.e. tree canopy
pruning. The overall treatments were:
2 tree row orientations * 2 plot orientation * 2 pruned/ unpruned * 2 distances
All measurements were done on all treatments (micro-plots).
The influence on crop growth was observed till maturity in a two-week interval.
Measurements included: crop height, the phenological stage (Zadoks scale) and the number
of organs (brown and green leaves, tillers). At each micro-plot, 5 plants representative for the
location were randomly selected. The measurement of flowering (onset, 50%, 100%) and
maturity (onset, 50%, 100%) were the most important stages.
In Vézénobres, around flowering we estimated leaf area and specific leaf weight of the upper
two leaves (leaf1 = flag leaf and leaf 2) of three plants of each treatment. The leaf area of
three representative plants was determined by measuring the leaf length and width (top,
middle and base). The length of the following leaves was measured too. As a calibration 30
plants outside the micro-plots were treated the same way and then harvested. The leaves
were photocopied and their leaf surface was measured by OPTIMAS. Then the leaves were
dried at 60 °C during 2 days and weighted for computation of specific leaf area. The specific
leaf area (m2 leaf kg-1 leaf) was computed based on the average leaf weight and estimated
leaf area. The analyses of the data are in work.
The number of all tillers were counted of half of the micro-plot (0.5 m2) in early May and the
number of panicles were counted of the entire micro-plot (1 m2) at the day of harvest.

In Mid-June, on DOY 200, the harvest took place about 229 days after sowing. The crop has
achieved physical maturity, but the drying of the grains was not finished at all measurement
locations. The number of difference in DOY when physical maturity was achieved was
visually measured in all micro-plots. The micro-plots plots were harvested manually (hand
clipper) after weeding. The plants were cut at ground level and the fresh weight of the
sample was measured. Then the number of panicles were counted and cut from the stem
and their fresh biomass was weighted too. For each micro-plot a sub-sample of 60 panicles
was randomly selected for grain weight, fresh and dry (48 hours at 60 oC).
Results
Climate and environmental site characterisation

The weather of the growing season 2003-2004 was rather dry as usual for the region ().
However, during the early crop establishment the plots were heavily flooded. In particular in
plot 1997 the water in the vicinity of several poplar rows was causing a delay in crop
development and at some places even death of plants. Nevertheless, we could find
reasonable sampling points.
100 60
Sowing Harvest
Daily rainfall (m m)
Daily m ax. temperature

80
Daily m in. temperature
40
Precipitation (mm)
Temperature ( oC)
60
20
40
0
20
0 -20
1 51 101 151 201 251 301 351 401 451 501 551 601 651 701
Time in day of year from start 2003 to end 2004
Table 12: Daily temperature and precipitation at Vézénobres in season 2003- 2004

40
Sowing Harvest
35
30
Daily solar radiation (MJ m )
-2
25
20
15
10
0
1 51 101 151 201 251 301 351 401 451 501 551 601 651 701
Day of year from start 2003 to end 2004
Table 13: Daily global radiation at Vézénobres in season 2003- 2004
Crop yields
In 2004, the grain yields of durum wheat in the poplar agroforestry stand were highly reduced
compared to the monocropping control plots (Table 8). Overall the reduction was about 50%,
with large differences between the treatments. At the first glance, two major effects can be
distinguished: pruning and orientation of the plots, while the distance to trees appears to be
less important (Figure 5).
Agroforestry treatment Yield in agroforestry (AF) Yield in Ratio

T/ha Trop control Yield AF/ Yield control
T/ha
All agroforestry plots 1.94 0.47 4.11
Unpruned Pruned Unpruned Pruned
Plot 96: Tree row N-S 2.08 2.38 5.16 0.40 0.46
Plot 97: Tree row W-E 0.97 2.35 3.06 0.31 0.77
Plot 97: Tree row W-E* 0.97 2.35 4.11* 0.24* 0.57*
** ** **
Plot 97: Tree row W-E 0.97 2.35 5.16 0.19 0.46**
* and **: Results using the mean of the two crop control plots and the value of Plot96, respectively.
Table 14: Durum wheat yields in an eight-year-old poplar stand at Vézénobres in 2004
It is striking that the production on the control plot 97 was in all years much lower than on plot
96. One of the reasons may be the shade of poplars in the morning (the control is located
west of the agroforestry plot). In dry regions, morning hours are very important with respect
to growth. The fact that the difference increased in 2004 (ratio yield plot96/ yield plot97 was
0.82 in 2003 and 0.61 in 2004) may also refers to the shade cause. When we correct the
yield of the monoculture crop plot 97 for an eventual shade effect (e.g. using the yield of plot
97 or the mean yield of both control plots), the influence of poplars appears to be higher on a
silvoarable field with West-East oriented tree rows (Table 7). Soil analysis will show if also
difference in soil fertility played a role.

8 8
SOUTH NORTH WEST EAST
6 pruned 6
élagué
unpruned
Yield (t.ha )
-1
non élagué
Yield (t.ha )
-1
crop control
4 4
2 2
0 -6 -2 2 6 0
-6 -2 2 6
Distance from poplar row (m) Distance from poplar row (m)
Table 15: Yields of durum wheat at different distances and orientations from a pruned
and unpruned poplar row in Vézénobres in 2004
The impact of the pruning regimes on wheat yield was impressive in the 1997 plot but less
pronounced in the 1996 plot (Table 15). The lowest yield was found in alleys between low-
pruned poplars, most striking in the south and north plots. Here the light condition is
heterogeneous and pruning treatment had the largest effect in the north, where light
reduction by low-pruned trees was highest (Figure 7). The standard errors (not presented)
are large, due to a combination of few repetitions and large spatial variability. In the East-
West tree rows plot, it must be noticed that the best yields were observed NORTH of the
poplars in 2004, which is the contrary to 2003. This can be explained by the simultaneous
effect if tree height increase and high pruning. Tree height increase moved the sunshade
further north, where it reached the next tree row, while high pruning allowed this light to
reach the zone situated North of the trees. This total change in only one year illustrates the
fast dynamics of a silvoarable system. The researchers advised the farmer not to seed the
zone north of the trees, but the farmers did seed. He was right!
From these data, is it possible to conclude that shade is the limiting factor for the wheat
production in this mature silvoarable system? Water competition may also play a role, as
pruning also reduces water use by the tree. However, unless we assume that the rooting
pattern of the poplars is not symmetrical on both sides of the tree row, the water competition
effect should be symmetrical. What we observe is a non-symmetrical impact well correlated
with the light availability that suggests that light is the limiting factor.
500 10000
SOUTH NORTH
SOUTH NORTH
400 8000
Number of grains.m ²
Number of panicle.m-²
300 6000
200 4000
pruned
élagué
100 2000 unpruned
non élagué
0 0
-6 -2 2 6 --6
6 -2
Distance from poplar row(m) -2 22 66
Distance from poplar row (m)
sud nord
Table 16: Number of panicles and grains of durum wheat on micro-plots at different
distances and orientations from a pruned and unpruned poplar row, Vézénobres in
2004

In plot 97, the higher yield on certain micro-plots can be explained by a higher number of
grains and an elevated “1000 grain weight” (Table 16). The higher number of grains is
mainly the result of the grain number per panicle, while the number of panicles is only
affected in the north of unpruned tree. The latter are the plots with the lowest light availability.
In plot 96 with generally more light (Table 17), there is almost no effect on panicle numbers
and grain weight (not presented), and the small effect occurs only in the vicinity of the trees.
Table 18 presents the correlation between light availability and grain yield and yield
components.
Wheat crop alleys between poplars Wheat crop alleys between poplars
pruned up to 6 m pruned up to 10 m
Table 17: Light availability, as % of global radiation transmitted to the crop, at different
orientation from the poplar row (green) at 2 m (yellow) and 6 m (orange) for both
pruning treatments Vézénobres in 2004
With the exception of DOY 120 (before booting) and the ripening phase, the phenological
development appeared to be similar everywhere on the silvoarable field as well as on the
crop control plot. Thus the influence of light availability had no effect on the wheat
development during most of its life cycle. At the more shady regions (micro-plots in the
vicinity of the north side of trees) the physiological maturity was delayed for about 7 days and
even at harvest, when wheat was mature everywhere in the agroforestry plots grains were
still more humid the more shady the micro-plots. Data are currently analysed.
Summary and conclusion

In general, the findings of the growing season 2003-2004 confirm the results of the prior
season. As in 2003, the influence of 7 m tall walnut trees on the winter crop was negligible,
while the reduction of durum wheat in the 20 m tall poplar plantation was remarkable.
The major conclusions that can be drawn are:
In Restinclières, the walnut trees are still to small to affect significantly the crop yield.
The late leafing date of the trees limit the shade impact, and their deep rooting
pattern limit the water competition
At Vézénobres, the mature poplar trees are at a very high density for an agroforestry
plot (156 trees/ha). Light competition is now strong, as a result of both early leafing

and large tree sizes. The poplars have a deep rooting pattern that also limits water
competition. Therefore, light competition seems to play a key role.
Canopy pruning of poplars was very effective for maintaining crop yields, due mainly
to shade reduction, but also possibly to a reduced water competition.
In 2003, root pruning had no effect on wheat yields, because trees used the water at
a deeper soil depth than the crop. Therefore the impact of root pruning on crop yield
was not measured in 2004.
4 500
y = 0.0417x - 0.052
R2 = 0.5892
400
Number of panicle.m ²
-
3
Grain yield (t.ha-1)
300
2
200
1 100
y = 1.1761x + 252.59
2
R = 0.1062
0
0 0 20 40 60
0 20 40 60 80 Fraction light transmitted to the crop (% )
Fraction light transmitted to the crop (% )
40
y = 0.3078x + 4.7173 50
R2 = 0.522
-1
Number of grain.panicle
30
40
1000 grain weight.g
30
20
20
10
10 y = 0.1106x + 26.399
2
R = 0.0538
0
0 20 40 60 80 0
Fraction light transmitted to the crop (%) 0 20 40 60
Fraction light transmitted to crop (% )
Table 18: Relation between light availability and yield components: grain yield,
number of grains per panicle, number of panicles per area, weight of 1000 grains
At Restinclières, by the time the walnut is developing its leaves (start and end of budburst
was DOY 110 and 156, respectively), the crop was already flowering and suffered only few
from shading. Similar competition for water and nitrogen are restricted to a short and less
important phase in the crop life cycle. In addition, the root studies showed that an important
part of tree roots are developed at much deeper soil depths than the zone occupied by
durum wheat roots.
At Vézénobres, the crop yield in the alley of poplar trees was hardly affected during the first
four years of the agroforestry system (i.e. 1997 –2000). Subsequently, with the increase of
the poplar canopy size, the grain yield was progressively reduced. In 2001, the reduction was
about 20% (3.7 t/ha AFS and 5 t/ha crop control), in 2003 and 2004 about 50%. In 2004, the
grain yield of durum wheat outside the influence of trees (average of control plots 4.1 t ha-1)
was higher than in the growing season of 2003 (average 3.4 t ha-1), partly due to better weed
control. Without additional pruning, the yields keep dropping in 2004: 40% in the 1997 plot,
30% in the 1997 plot. But the additional pruning was very effective in increasing the crop
yield to 50% in the 1996 and 70% in the 1997 plots.

In 2004, the distance to the poplar row had a minor role on crop production. In the last year
the poplar trees were tall with large (partly each other touching) canopies, hence light was
more homogenous distributed over the transect of an alley.
Water competition cannot be excluded, but the results suggest that in view of competition
radiation was predominantly affecting the yield.
WP4 Above-ground interactions

Improving the Hi-sAFe light competition module (INRA-System)
Field observations show that branches of mature deciduous (without leaves) trees produce
significant shade in winter. This reduction of light can affect winter crop physiology. A simple
way to account for the branch shade was to assume that a low leaf area density was still
present in winter. This has been added to the Hi-sAFe phenology module. To simulate trunk
and branches shade of trees before bud burst and after leaf fall, field measurements are
necessary (hemispheric pictures) to get a value for the new parameter that is called
winterVirtualLad with a default value of 0.03 m2 m-3. However, this modification implies that
the light module is now executed even in winter season, which significantly slow the model
runs.
Table 19: Hi-sAFe outputs showing how a 8 meter high walnut tree significantly
shades the crop at noon, on 21st December (2 months after leaves fall)
While testing, it appeared that the Hi-sAFe light module did not predict correctly the
interception of direct light by the tree on the scene when both the cells are small and the
scene is large compared to the canopy size. The problem is more serious when the tree
canopies are high and narrow, such as for poplars. This was the consequence of direct beam
interception being calculated only five times a day. In that case, some cells escaped the
direct shade of the canopy, and this resulted in incorrect direct beam maps on the scene
(Table 20, top).
The solution consisted in running the direct interception routine more frequently each day. In
the last version of Hi-sAFe, the number of calculations per day is a parameter that the user
can tune in the range 5 to 11. A further improvement would be an automatic optimisation of
this parameter by taking into account the size of the cells, the width of the canopy, the height
of the canopy and the height of the pruned part of the tree.
However, it can be checked on Table 20 (bottom) that even with 11 calculations per day,
some anomalies still exist at a distance of the tree canopy larger than two diameters of the
canopy. Projected shades of the canopy are still disjoined, resulting in direct shading being
underestimated for some cells.

Table 20: Hi-sAFe prediction of the daily integrated tree shade, the 15th of May, with
different frequencies of direct beams interception: 5 times a day above and 11 below
Note that the integrated daily radiation map exhibits a banana-shaped shade that is very
close to the a priori shape that we used in the further steps of the design of the Hi-sAFe
model.
Table 21: Assumption of the integrated daily shade of an isolated tree as hypothesised
during the WP1 work on the Hi-sAFe concept (The North is in the right top corner).
WP5 Below-ground interactions

Implementing the two concurrent water extraction modules
During the last 6 months, the two water extraction modules (INRA and ICRAF versions) were
implemented in the Hi-sAFe shell and evaluated. Both are now correctly operating. The full
descriptions of the modules are already available in deliverable 5.1 for the INRA module, and
in the ICRAF final report for the ICRAF module.

Exploring the behaviour of the integrated model
The water and root routines of the model were tested using the 2002 and 2003 weather data
at the Restinclières experimental plot near Montpellier (Table 22).
180 2500
160 Daily rain

Cumulated rain
2000
140
120
1500
100
80
1000
60
40
500
20
0 0
01/01/02 01/07/02 01/01/03 01/07/03 01/01/04
Table 22: The rainfall pattern during the years 2002 and 2003 was similar with very
heavy rains in autumn that caused floods in the area.
Both modules predict a strong water stress during the second year (Table 23). The INRA
module however indicates that the tree will not be able to extract water during 2 months
(which means that the tree would die). Therefore, the module based on the water potential of
the plants appears to be more efficient.

200
waterDemand
180 waterUptake
160
140
120
-1
Using the ICRAF water

L day
100
80
competition module
60
40
20
0
01/01/02 01/04/02 30/06/02 28/09/02 27/12/02 26/03/03 24/06/03 22/09/03
200
waterDemand
180
waterUptake
160
140
120
L day-1
100 Using the INRA

80
competition module
60
40
20
0
01/01/02 01/04/02 30/06/02 28/09/02 27/12/02 26/03/03 24/06/03 22/09/03
Table 23: The tree water demand and uptake of a 10 year old walnut tree predicted by
the Hi-sAFe model at Restinclières in 2002 and 2003.
The pattern of prediction of soil evaporation, crop transpiration and tree transpiration was
satisfactory (Table 24). The comparison with field data is still in the making, and will be
performed during the next year as part of the validation of the model, as explained in the
Technological Implementation Plan.
Environmental indicators such as deep drainage of water and nitrogen under the tree stands
can be calculated using the model (Table 25). A preliminary assessment shows that the
model predicts a significant decrease of water drainage under a plot of 10-year-old walnut
trees at a density of 100 trees per hectare. Such trees are still small and their root systems
do not explore the whole are of the scene. Much higher impacts can be expected with larger
trees. The consequence on nitrogen leaching was not yet quantified, but should be
significant.

7
Tree transpiration
6 Crop transpiration
Soil evaporation
5
mm day-1
0
01/01/2002 01/04/2002 30/06/2002 28/09/2002 27/12/2002 27/03/2003 25/06/2003 23/09/2003
Table 24: Soil evaporation, crop and tree transpiration of the agroforestry walnut plot
at Restinclières as predicted by the Hi-sAFe model.
900
P u re c ro p
800
AF
700 T r e e c e ll in A F
600
500
mm
400
300
200
100
0
0 1 /0 1 /0 2 0 1 /0 4 /0 2 0 1 /0 7 /0 2 0 1 /1 0 /0 2 0 1 /0 1 /0 3 0 1 /0 4 /0 3 0 1 /0 7 /0 3 0 1 /1 0 /0 3
Table 25: Deep drainage as predicted by Hi-sAFe during the very rainy 2002-2003
winter at Montpellier. The tree component of the agroforestry plot reduces the
drainage by half in the tree-rooted cells, and by 16% on the whole agroforestry plot.
No runoff was assumed for this calculation.

35%
30%
25%
20%
Tree cell in the
15% AF plot
0,1
10% 0,6
1,5
2,5
5%
3,5
0%
01/01/02 01/04/02 01/07/02 01/10/02 01/01/03 01/04/03 01/07/03 01/10/03
35%
30%
25%
20%
Pure crop
15%
0,1
10% 0,6
1,5
2,5
5%
3,5
0%
01/01/02 01/04/02 01/07/02 01/10/02 01/01/03 01/04/03 01/07/03 01/10/03
Table 26: Soil water moisture dynamics as predicted by Hi-sAFe at the Restinclières
agroforestry plots
The model appears to simulate correctly the soil moisture heterogeneity (Table 26). Soil
moistures dynamics are quite similar in the topsoil in the cropped area and the tree row, and
are dominated by soil evaporation. However, the crop water extraction is limited to the first
meter of soil, while the tree extraction is much more deeper. The root voxel automaton
correctly predicts the distortion of the tree root system that is expanding faster in deeper and
moist soil horizons (Table 27). For these simulations, the fine root density threshold for
colonisation of neighbouring voxels was fixed at 500 m m-3. This value needs further
adjustment using field data. We initialised the rooting pattern of the tree with a uniform
density in the rooted zone, and the model predicts a reshaping of the root system that is
close to the measured profiles in the experimental fields.

1000
0,1
900
0,6
800 1,5
2,5
700
3,5
600
M m-3
500 Tree cell in the

400 AF plot
300
200
100
0
01/01/02 01/07/02 01/01/03 01/07/03
1000
0,1
900
0,6
800 1,5
2,5
700
3,5
600
First cropped
M m-3
500
cell next to the
400
tree
300
200
100
0
01/01/02 01/07/02 01/01/03 01/07/03
1000
0,1
900
0,6
800 1,5
2,5
700
3,5
600
Second
M m-3
500
cropped cell
400
next to the tree
300
200
100
0
01/01/02 01/07/02 01/01/03 01/07/03
The tree fine roots were initialised with a uniform fine root density in a spherical rooted
volume. The root colonisation threshold was 500 m m-3 in these runs. The distortion of the
rooted profiles is clear.
Table 27: Tree fine roots dynamics in the agroforestry plot at Restinclières
As a conclusion, the belowground modules of Hi-sAFe for root growth and water extraction
appear to behave correctly, but need further validation by comparison with the field data.
This is the priority for the INRA team after the SAFE project
Implementing a Nitrogen competition module in Hi-sAFe
A nitrogen module, linked to the ICRAF water competition module, has been delivered in
January 2005 by the ICRAF sub-contractor. A word document describes the module
algorithm and an excel sheet provides the equations. The full documentation on the Nitrogen
module is available on line of the SAFE web site.

Basic principles
From single root to root system
De Willigen et al. (2000) discussed how models at the level of root systems in a volume of
soil could use equations that were primarily derived for the concentration profiles around
individual roots, using a steady-rate solution to the equations describing diffusion in a
cylindrical co-ordinate system. If plant demand for nutrients is high most plants can exhibit a
physiological uptake efficiency that is so high that we can describe roots as ‘zero sinks’,
maintaining a concentration of virtually zero at the root surface. The amount of nutrients that
can reach the root surface then depends on the effective diffusion rate (influences by soil
water content), the average concentration in soil solution (that is related to the ‘available’
amount via a sorption constant) and the geometry of the system with the root surface area as
inner diameter and the midpoint between roots as the outer boundary. An explicit equation
describes the potential uptake supply as a linear factorial of diffusion and concentration and
a non-linear function of root length density (Fig. 1). The basic equation has been used in a
number of models describing the dynamics of nutrient and water uptake by monocultures (De
Willigen and van Noordwijk 1987, 1994, 1995; Heinen, 2001).
0.25 Root diameter (cm) 0.6

0.02 Soil water content, vol/vol
0.20 0.5
Uptake options
Uptake options
0.05 0.15
0.4
0.15
0.20
0.3
0.10 0.30
0.2
0.05 0.1
0.00 0.0
0 2 4 6 8 10 0 2 4 6 8 10
Root length density, cm /cm 3 Root length density, cm /cm 3
Table 28: Relationship between potential uptake during a 1-day time step from a
volume of soil and root length density; A. for two values of root diameter; B. for three
levels of soil water content
A related equation describes the potential supply to a root system by a combination of mass
flow and diffusion – because the two processes show strong interaction and the combined
result differs substantially from the sum of both processes. Quantitatively mass flow is more
important for nutrients with relatively high concentrations in soil solution (such as nitrate) than
it is fore nutrients with low concentrations (such as phosphate). But a comparison of the total
supply with or without mass flow shows, by contrast, that mass flow has a larger positive
effect on possible uptake rates by the root for phosphate than it has for nitrate. The error
made by ignoring the process of mass flow is probably of the same order of magnitude, but
opposite sign of the error made by describing roots as zero sinks, rather than in need of
maintaining a finite minimum concentration at the root surface area.

0
Error by ignoring massflow

-0.05
-0.1
-0.15
-0.2 0.5 cm day-1

0.3
-0.25
0.1
-0.3
0.01 0.1 1 10
-3
Root length density, cm cm
Table 29: Relative error in the calculated potential uptake rate by a root system of
given root length density from a unit volume of soil if the contribution of mass flow
(dependent on actual water uptake) is ignored, on the basis of the equations
developed by D
The switch (back and forth..) between supply and demand-limited situations can be made
easily once a quantitative method exists for estimating daily demand given the current
biomass, growth rate and nutrient content. By applying the lowest of the supply and the
possible supply, the nutrient contents of the crop and plant can be updated for the next time
step. Where total possible supply exceeds the current demand, actual uptake from any voxel
of soil can be taken as a proportional share of the total – thus approaching a ‘minimum
energy extended’ concept. The strong role for ‘plant demand’ in down-regulating uptake is in
line with the plant physiological literature of the past decades, that has superseded the
earlier focus on concentration-dependent uptake (and the related modelling approaches of
the Barber school).
In the case of nitrogen most plants deal with two inorganic forms: ammonium and nitrate.
The ratio at which they contribute to the ‘soil mineral N’ pool depends on the pH, activity of
nitrates and recent fertiliser history. On most soils under agronomic use nitrification is rapid
and nitrate will dominate. As there is good evidence that the plant regulates the combined
uptake of the two N forms rather than ammonium and nitrate uptake separately, we can sum
the potential uptake from both sources – acknowledging a substantial difference in
adsorption behaviour. As the equation for the potential supply to a zero sink is linear in the
concentration term, we can add the potential ammonium and nitrate supply to a ‘mineral N’
heading by adjusting the apparent sorption constant (that relates concentration in soil
solution to the available stock).
From mono-specific to mixed species vegetation
The basic equations can be applied to root systems in a volume of soil by first adding up the
roots of all species present to a total root length density with its ensuing potential zero-sink
nutrient supply, and then sharing out the potential uptake over the various species
proportional to their contributions to total root length density. In doing so a number of
additional considerations are:
- differences in root diameter between the plant species: this can be done by
introducing the concept of ‘weighted mean root diameter’
- differences in current demand that may make the ‘zero sink’ assumption an
overestimate for some of the plants; by using the current demand per unit root length

at plant level as additional weighing factor roots of plants with no current demand
don’t influence the possible uptake by roots of plants with a strong demand
- the possible uptake by any plant in a mixture can not be more than what it could get
in a monoculture for the same root length density and soil concentration; a series of
constraints on the uptake sharing rules ensure that adding roots of a non-demanding
plant to the total root length density will not increase the possible uptake by the
others…
With these additional rules, an efficient algorithm can be constructed. In the translation from
the WaNuLCAS routines to Hi-sAFe, a number of minor changes were made.
References
de Willigen, P. and van Noordwijk, M., 1987. Uptake potential of non-regularly distributed
roots. Journal of Plant Nutrition 10: 1273-1280.
De Willigen, P. and Van Noordwijk, M., 1994. Diffusion and mass flow to a root with constant
nutrient demand or behaving as a zero-sink. II. Zero-sink uptake. Soil Science 157:
171-175
deWilligen, P. and van Noordwijk, M., 1995. Model for interactions between water and
nutrient uptake. In: Kabat, P., Marshall, B., Broek, B. J. van den, Vos, J., Keulen, H. van
(Eds.) Modelling and Parameterisation of the Soil-Plant-Atmosphere System: a
Comparison of Potato Growth Models. Wageningen Pers. Wageningen pp 135-153
de Willigen, P, Nielsen, NE, Claasen, N and Castrignano, AM. 2000. Modelling Water and
Nutrient Uptake. In: Smit, A.L., Bengough, A.G., Engels, C., van Noordwijk, M., Pellerin,
S. and van der Geijn, S. (Eds.) Root Methods, a Handbook. Springer Verlag, Berlin. Pp
509-544.
Gregory, PJ. 1996. Approaches to modelling the uptake of water and nutrient in agroforestry
systems. Agroforestry Systems 34: 51- 65.
Heinen, M. 2001. FUSSIM2: brief description of the simulation model and application to
fertigation scenarios. Agronomie 21: 285 – 296
Specific experimental protocols for validating the root module of Hi-sAFe
Objective
The root voxel automaton is an important aspect of the Hi-sAFe model. It includes 6
parameters that describe the sensitivity of root systems to soil heterogeneity. To calibrate
these parameters, container experiments were set up with hybrid walnut and wild cherry
trees grown in containers. The final assessment of the experiment was made in November
and December 2004 by excavating the trees and measuring the fine root densities. We
report here these results. The use of these data for calculating the parameters for Hi-sAFe is
still under progress and cannot be reported here. It will be part of the Ph. D. thesis of
Rachmat Mulia who is due to defend his thesis in May 2005.
These container experiments were presented in the previous SAFE reports. They aim at
observing the growth of tree root systems in heterogeneous soil conditions. This is useful to
better understand and predict the plant rooting behaviour in natural environments, which are
spatially (and temporally) patchy in terms of soil resources. The root data will also be used
to validate the root voxel automata (RVA) model specially designed to fully take into account
the local soil resource conditions when simulating plant root growth.

Materials and Methods
The experiments were carried out at INRA Montpellier (Figure 1 A). We used two temperate
tree species: hybrid walnut (Juglans hindsii * regia) and wild cherry (Prunus avium) which
have some difference related to their aboveground and belowground physical characteristics.
For example, the first has compound leaf type, relatively low ratio between canopy height
and width, and a vertical main root (a pivot), which are not the case with the second. The
walnut also has a shorter growing period than wild cherry. The tree materials (one-year old
walnuts and wild cherries) were bought from the Peyre tree nursery near Grenoble.
Each tree was planted in a container with 60 cm and 50 cm of top and bottom diameter
respectively, and 50 cm height (i.e. 120 L volume). Before planting, the structural root system
of each selected tree was shaped to have 10 cm x 10 cm width and 25 cm depth. Only a few
numbers of living fine roots (diameter < 2 mm) were left. In the aboveground system, all trees
were also homogenised several weeks after bud burst. Generally, each walnut was left with
only two long shoots and each wild cherry with four short and two long shoots. Each pot was
basically divided into four horizontal layers (i.e. each of 12.5 cm thickness) separated by
plastic grids (with 2 cm x 2 cm mesh size) and each tree was planted in the middle of the two
first horizontal layers (Figure 1B).
A) B
d=60 cm
Plastic
h=50 grids
cm
h=12.5
cm
d=50 cm
Table 30. A) The pot experiments carried out in the year 2004 to observe the growth of
plant root system in a heterogeneous soil condition. B) The dimension and
configuration of each pot used in the experiments.
The experimental design aimed at measuring the effect of soil water and nutrient
heterogeneity on the growth of the tree root system. Basically, we introduced the resource
heterogeneity in two ways: either enriched soil resource was localised in a small area inside
pot or different levels of soil water and nutrient content were set among horizontal layers or
vertical compartments. Two kinds of substrate were used: soil from the Restinclières
experimental plot (25.5 % of clay, 59.5 % of silt, 15.0 % of sand, 1.85 % of organic matter,
1.07 % total C, 0.85 % total N, and 12.6 C/N ratio); and perlite (expanded clay) which
contains no nutritive elements (Morel et al., 2000).
The type and detail of each experiment are described below. The first two types were related
to soil water whereas the others were concerned with nutrient heterogeneity.
Experiment 1: Pot experiment with localised soil water. We used soil substrate and a small
soil area (10 cm x 20 cm with 12.5 cm thickness) in the third horizontal layer was continually
irrigated. In contrast, no irrigation at all was applied to the other soil areas. During 5 months
(May – September 2004), an automatic dripper (Tropf Blumat system, Two Wests & Elliott
company, Derbyshire, UK) was used as an irrigation system (Figure 2).

Table 31. The system of irrigation used in the pot experiment with a localised soil
water supply (Copyright Two Wests & Elliott company).
The probe was filled with water and the section with ceramic material was located inside the
irrigated soil area. However, we actually used the probe with a much longer water-filled tube
(i.e. the section between the ceramic probe and the dripper head) so we could easily
manage the setting knob from the soil surface. The probe type described in Figure 2 is the
short version. Basically, when the soil near the probe position dries then water starts to be
sucked out of the porous ceramic probe causing a diaphragm to be pulled downwards. This
allows water to flow through the drip tubing. We placed two drip tubes (with 10 cm distance)
on the ‘surface’ of the third layer (i.e. 25 cm depth) to evenly humidify the irrigated soil area.
As the soil become more humid then the water suction is reduced causing the diaphragm to
move upwards restricting the water flow.
Experiment 2: Pot experiment with different soil water content among horizontal layers. We
used gravels mixed with soils (1:1 in volume) and pure soils. The mixed substrates were
located in the top and the third layer and pure soil in the second and in the bottom layer.
Gravels reduced the fine soil matrix in the mixed layers, reducing the water holding capacity
per volume by half. Hence, there would be differences of soil water content among the mixed
and pure soil layers that may give an effect to plant root growth.
Experiment 3: Pot experiment with localised nutrient. The principle is the same as
Experiment 1 but here we used perlite instead of soils. Two voxels (10 cm x 20 cm with 12.5
cm thickness) in the third horizontal layer were enriched with nutriments in the form of
granules (Osmocote, 14% N: 13% P: 13% K).
Experiment 4: Pot experiment with different nutrient content among horizontal layers. We
used soils in the top and the third layer and perlite in the other two layers. This result in a
contrast for nutriment availability.
Experiment 5: Pot experiment with different nutrient content between two vertical
compartments. Basically, each pot was still divided into four horizontal layers but a half of pot
contained pure soils and only perlite in another (Figure 3A).
Experiment 6: Pot experiment with different nutrient content among horizontal layers and
used to verify the ability of roots to go upwards. Basically, each tree was planted in the
middle of the first two soil layers as described before, but the root system was surrounded by
a plastic root barrier (10 cm * 10 cm, and 25 cm height). Therefore, all tree roots grew first in
the third layer (i.e. the height of the barrier corresponds to the depth of two first horizontal
layers). Any roots measured in the first or second layer would result from a negative
geotropism. This experiment consists of two types: in the first type, the two first horizontal
layers contained soils and only perlite in the two deeper layers. We would verify if this
situation can attract roots to go upwards (i.e. into the first two layers). As a control (the
second type), we used only perlite in all layers.

All trees were watered (with a watering can) two or three times a week depending on the
rainfall occurrence. However, for the trees used in the localised soil water experiment, no
manual watering was carried out after May 2004. Specifically for the experiment with only
perlite substrate used, we also watered trees with one litre solution nutritive every week
evenly distributed on pot surface.
From October to December 2004 (when growing season has completely terminated), two
trees of each species of each experiment type were progressively uprooted. However,
because a number of wild cherry trees died due to a delayed plantation time then there was
only one wild cherry or only walnuts available for certain experiment types. The trees were
uprooted either voxel per voxel or layer per layer. The compartmentalisation of each layer
into 21 voxels (each of 10 cm x 10 cm and 12.5 cm thickness) and voxel numbering is
described in Figure 3B. The trees used in the experiment with localised soil resource and in
the experiment to verify the ability of roots to go upwards were uprooted voxel per voxel;
otherwise, roots were collected layer per layer. Two types of tree root were distinguished:
fine (diameter < 2 mm) and structural root (diameter ≥ 2 mm) but only data of fine root will be
presented here.
A) B)
North
13 14 15 10 cm
12 3 4 5 16 10 cm
West 11 2 1 6 17 East
10 9 8 7 18
21 20 19
South
Table 32. A) Pot experiment with different nutrient content between two vertical
compartments. One compartment contained fertile soils and only perlite in another. B)
Voxel numbering.
Results
Experiment with localised soil water supply

The distributions of fine root biomass (g dm-3) of walnut and wild cherry used in the
experiment of localised soil water are described in Figure 4. The description was done for
each horizontal layer with the SURFER software (Golden Software Inc, Colorado, USA). The
circle in the first figure was made to remind that we collected roots in 21 voxels of each layer
and the root biomasses near pot border were not used in the analysis.
The heterogeneous soil water condition strongly affected the growth of the walnut root
system (Figure 4A). Roots were more concentrated in the areas close to the irrigated soils
and this situation can be regarded in the second, the third, and the last soil layer. However,
the highest root concentration was indeed found in the last instead of the third soil layer. This
fact likely indicates that water largely infiltrated to the bottom of the pot which makes the soil
area directly beneath the probe position was very favourable for root growth.

For wild cherry, the experiment is not satisfactory. We had no replication : the second tree
did not survive. And the reliability of the water distribution is questionable. In July 2004 we
found that the plastic drip tube leaked (due to a wasp sting) and the pot was flooded. This
accident might have contributed to absence of effect of the localised soil water supply on the
tree root distribution (Figure 4B).
A) Hybrid Walnut (two replications)

g dm-3
1.6
1.4
1.2
1
T 0.8
0.6
0.4
0.2
0
B) Wild Cherry (no replication; water distribution not reliable (see text))
Top layer Second layer Third layer Bottom layer
The letter ‘T’ indicates plant position. The black rectangle indicates the position of the
localised water supply
Table 33. Distribution of fine root biomass (g dm-3) of walnut (A) and wild cherry (B)
inside pot with localised soil water supply.
Experiment with localised nutrient supply

The effect of localised nutrient supply on the root system of walnut (Figure 5A) and wild
cherry (Figure 5B) was impressive (look at the picture on the cover page of this report). In the
third perlite layer, roots were clearly more concentrated inside the nutrient-enriched areas. It
is worth to remind here that we watered the trees with one litre of nutritive solution every
week. Therefore, roots may basically grow well in other perlite areas as well.

A) Hybrid Walnut (2 replicates) g dm-3
1.6
1.4
1.2
1
T 0.8
0.6
0.4
0.2
0
B) Wild cherry (no replicate)
Upper layer Second layer Third layer Bottom layer
The letter ‘T’ indicates plant position. The black rectangle indicates position of
localised nutrient supply.
Table 34. The distribution of fine root biomass (g dm-3) of walnut (A) and wild cherry
(B) inside pot with localised nutrient supply.
Intriguingly, we found that overall root distribution pattern of walnut and wild cherry was
different. The bulk of wild cherry roots were found in the top 12.5 cm perlite layer and it was
not the case with walnut root system. In the latter, the roots were more developed in the
second layer with relatively higher concentration in deeper perlite layer. Related to the
number of tree uprooted, we had two walnuts but only one wild cherry was available.
Experiment to verify the ability of roots going upwards

As mentioned earlier, the experiment to verify the ability of roots to go upwards (negative
geotropism) consists of two types: in the first, we used only perlite so the growing condition
inside pot was homogenous but in the second, the first two layers contained fertile soils
instead of perlite so there was a strong difference of nutrient content between the upper and
deeper layers. This experiment included only walnut trees.
The distribution of walnut root density (g dm-3) inside pot with homogeneous and
heterogeneous growing condition is described in Figure 6A and Figure 6B respectively. In
both situations, roots of walnut were able to grow upwards and were found in the first two
layers. Interestingly, the deeper perlite layers were clearly not fully occupied which indicates
that the decision of roots to go upwards was not related to root saturation or space
occupation in deeper layers.
It is again worth to remind here that we watered trees in pots of pure perlite with one litre
nutritive solution every week. Due to this, the fact that roots go upwards may be actually due
to genetic rooting behaviour or nutrient supply in the first two upper layers. However, it is
clear that if the upper layers are really attractive in term of nutrient resource then relatively
less roots would be found in deeper layers and more roots would go upwards as can be seen
when comparing Figure 6A and 6B.

A)Perlite substrate in the whole container g dm-3
1.6
1.4
1.2
1
T 0.8
0.6
0.4
0.2
0
B)Soil substrate in the two top layers and perlite in the two bottom layers
Upper layer Second layer Third layer Bottom layer
The letter ‘T’ indicates plant position. The black rectangle indicates position of
root barrier.
Table 35. The distribution of walnut fine root biomass (g dm-3) of walnut inside pot
used to verify the ability of roots to go upwards.
Pot experiment with a horizontal heterogeneity

In this experiment, tree root systems were uprooted per layer. Therefore, we had only two
root density values for each layer correspond to that in the soil and perlite compartment. We
uprooted two walnuts and two wild cherries. It should be recalled that we introduced no
nutritive elements in the perlite compartments.
The root density of walnut (Figure 7A) and wild cherry (Figure 7B) was higher (based on one
standard error) in the soil than in the perlite substrate. Nonetheless, it may not be surprising
since the soil is much more favourable than perlite in term of nutrient availability. Related to
water supply, we watered each tree two or three times a week (as mentioned before) so the
effect of different water resource between the two types of compartment may not be
expected here.
Another interesting aspect observed in Figure 7 is the root distribution with depth. It is
evident that the walnut and wild cherry had a different root distribution pattern. For wild
cherry, root concentration was higher in the first layer than that in deeper layers but it was
not the case with the walnut. For the latter, there is a tendency that root density was
increasing with depth and this took place both in the soil and in the perlite compartment.

A) Hybrid walnut B) Wild Cherry
1,5 1,5
fine root density (g dm -3)

Soil
Soil
Perlite
1 Perlite 1
0,5 0,5
0
0
0 10 20 30 40 50
0 10 20 30 40 50
depth (cm) depth (cm)
Table 36. The distribution of fine root biomass (g dm-3) of walnut (A) and wild cherry
(B) with depth, inside pots with horizontal heterogeneity. The vertical bars indicate
one standard error.
Experiment with different nutrient or water content among horizontal layers

The distribution of tree root biomass (g dm-3) with depth is shown in Figure 8A and Figure 8B
for the experiment with different soil water content (i.e. Experiment 2) and nutrient content
(i.e. Experiment 4) among horizontal layers. We uprooted two walnuts and two wild cherries.
A) gravel /soil / gravel / soil layers B) Soil / perlite / soil / perlite layers
1,5 1,5
Walnut Walnut
WdCherry WdCherry
1 1
0,5 0,5
0 0
0 10 20 30 40 50 0 10 20 30 40 50
depth (cm) depth (cm)
The vertical bars indicate one standard error.
Table 37. Distribution of fine root biomass for young trees grown in containers with 4
contrasted layers.
As described earlier, to create a heterogeneous soil water condition inside pot, we used
gravels mixed with soils (1:1) in the first and the third layer and soil only in the two other
layers. If this situation happens then root concentration would be a priori higher in the second
and the last horizontal layer. However, we did not control soil humidity of each layer.
Gravel layers did prevent root growth significantly. This is very impressive for wild cherry
(Table 37A) : roots in the top gravel layer were very scarce (compare with Table 36B). These
poor growth conditions in the top layer seem to have affected the whole wild cherry tree

growth. For walnut, the impact is impressive in the third layer (less roots than in the second
one, which is the opposite of the pattern evidenced in Table 36A.
For the experiment with different nutrient content among horizontal layers, no difference
between root density in perlite and soil layers was surprisingly observed (Figure 8B). Only for
wild cherry, we observed that root concentration in the top layer was much higher than that in
the second layer containing perlite. However, this may not indicate that a clear evidence of
nutrient effect was found because high root density in the top layer is generally observed for
wild cherry as described in the previous figures.
Conclusions
These are preliminary and fresh results of these experiments. We regret that we cannot give
a full account at the time of reporting. However, conclusions for the Hi-sAFe model are as
follows:
Tree root systems react rapidly to soil heterogeneity: this was the central hypothesis for
designing the root module of Hi-sAFe. It is confirmed.
Both tree species react similarly to soil heterogeneity. This result in a distortion of the
reference rooting pattern that they exhibit in a homogeneous soil. We already knew that
walnut and wild cherry rooting systems behave very differently in a homogeneous soil (see
previous reports). Therefore, the analysis of the results should be done by calculating the
deviation from the reference patterns on a homogeneous soil.
We may therefore hypothesize that the same parameters will correctly describe the root
sensitivity to soil heterogeneity for the two tree species. This will complete the genetic
parameters that are different for each species (see previous reports).
Global vertical and horizontal heterogeneity have very different impacts on tree root systems.
• Horizontal heterogeneity is more effective in distorting root growth (compare Table 36

and Table 37). A favourable soil zone attracted tree roots (with about twice as many roots in
the fertile soil as compared to the perlite zone for both tree species). This is very important
for agroforestry, because agroforestry is the only system where horizontal heterogeneity is
systematically occurring due to the patterns of extraction by the different neighbouring
species.
• Vertical heterogeneity has less impact on the root distribution: unfavourable layers are
less colonized, but the whole profile remains similar to the reference pattern. This is common
in most soils and cropping systems.
It is difficult to compare the relative strength of the water and nutriment heterogeneity with
our experiments. The most impressive distortion was obtained with the water heterogeneity
for walnut and for the nutriment heterogeneity for wild cherry
Finally, we must admit that for quantifying the reaction of tree root systems to soil
heterogeneity, more intensive monitoring of soil variables would be necessary in 3D, and this
was not within our possibilities during this project.
WP6a Biophysical integrated plot model Hi-sAFe

INRA-SYSTEM is responsible for the implementation of the Hi-sAFe model under the
CAPSIS 4 environment. I. Lecomte is in charge of this aspect, which is the main activity of
INRA-SYSTEM for WP6.

STICS adaptation checking
A lot of tests have been performed to check if STICS adaptation gave correct results, after
linking with both water repartition modules. These test have been made with maize and
durum-wheat STICS crop species parameters. The most important modification to check was
the Hi-sAFe voxels to STICS mini-layers aggregation and desegregation procedures,
computed every day, before and after each STICS interruption, and for each cell of the
simulated plot.
START Read data inputs Tree Crop

(soil-tree-crop) initialisation initialisation
Tree phonology
Crop re- Read daily

END initialisation weather inputs
Tree management
YES
NO YES
Last Is leaf area
rotation ? >0?
YES NO Light interception

Rain interception
NO Tree water demand
Last day ?
Light interception
Rain interception
Stemflow
Water and crop roots density

in STICS mini-layers are Crop phenology
desegregated in each voxels
Crop growth
Crop management
Light interception
Crop water demand
Crop stress indexes Crop water

calculation extraction and stress Crop water demand
indexes calculation Is tree water
demand
YES
NO >0?
Is tree water
extracted
NO Tree and crop water extraction YES
>0? in voxels are aggregated in
each STICS mini-layers
Water competition and soil
water extraction calculation
Crop root growth Soil water and nitrogen C allocation
(trees and crops)
tranfert Root growth
Table 38: Hi-sAFe daily loop showing voxel mini-layers aggregation and
desegregation procedures (red arrows)
Another important point to check was crop rotation chaining on several years. STICS is
initially designed to compute single crop season simulation. Several years’ simulations are
nevertheless possible but implement a data copy and paste procedure on an ASCII file
(stored on the hard disk of the computer) between each year. Soil state variables are written
on this file at the end of one year, and are read at the beginning of the following year to
reload the state of the ground. To avoid this heavy and time-consuming operation, the
chaining year procedure has been totally reviewed (so called crop re-initialisation in the
figure above).
A summary of the STICS model modifications that have been required for inclusion in
the Hi-sAFe model
Breaking the daily loop of STICS

La boucle annuelle de STICS a été supprimée afin de ne garder que la boucle journalière. Le
programme résultant a été compilé sous forme de DLL afin d’être piloté depuis Hi-sAFe, au
jour le jour, sur chaque cellule indépendante constituant la scène à simuler.

Table 39 : Scène de simulation Hi-sAFe ou chaque cellule carrée
correspond à une instance de STICS (vert=maïs, blanc=sol nu)
Grâce à une intervention de Dominique RIPOCHE en automne 2002, les variables
auparavant regroupés dans un seul common (paquet de variables globales) ont été
séparées en paquets distincts :
• Les paramètres généraux et sol (Y4)
• Les paramètres plantes et itinéraires techniques (Y8)
• Les variables de sortie (Y6)
• Les variables climatiques (Y3)
• Toutes les autres variables simples (Y1)
• Toutes les autres variables sous forme de tableaux de nombres réels (Y7)
• Variables relatives aux cultures associées (Y2 et Y5)
Des routines C permettant la sauvegarde/restauration des tampons mémoire Y1 et Y7 ont

été écrite dans STICS afin de conserver les valeurs des variables d’état des cellules (ou
instance de DLL) d’un jour sur l’autre.
Les tampons Y3, Y4 et Y8 n’ont pas à être sauvegardés car ils correspondent à des
paramètres stables pour une rotation culturale. Le tampon Y6 n’est pas utile dans Hi-sAFe
qui gère lui même ses sorties. Les tampons Y2 et Y5 ne sont pas non plus utiles car on ne
désire pas planter 2 cultures dans la même cellule.
Fichiers paramètres
La plupart des fichiers paramètres lu dans STICS sont redondants avec ceux déjà utiles à
Hi-sAFe. Les routines de lecture ont donc été déconnectées de STICS puis remplacées par
le passage de tableaux mémoire reçu de Hi-sAFe
• Paramètres du sol (param.sol)
• Entrées climatiques (.sta)
• Itinéraires techniques (.tec)

• Paramètres de simulation et valeurs initiales (.usm)
Seules les routines de lecture des fichiers param.par et plante.plt restent dans STICS.
Des interfaces JAVA ont été écrite pour piloter la simulation (nbr jours, espèces, itinéraires
techniques, fichiers climat) et pour gérer les itinéraires techniques de la culture.
Table 40 : Interface JAVA permettant le lancement d'une simulation
Table 41 : Interface JAVA pour gestion des itinéraires techniques
Improving the code

Au cours de différents test, il s’est avéré que certaines variables n’étaient pas bien
positionnées dans les différents tampons, et étaient donc écrasées d’un jour à l’autre lors
des copier/coller des tampons. Ces variables ont été renommés correctement dans le code
afin de rétablir la conformité des résultats (tests effectués sur blé, mais, soja, colza et sol
nu). De plus, certaines partie de code ont été nettoyé pour éliminer des variables inutiles ou
gênantes (variables textuelles remplacées par des variables entières, test d’option inutiles à
Hi-sAFe etc…) Chaque partie de code modifiée a été repéré grâce à un commentaire
spécifique (//MODIF SAFE )

Chaining years
La procédure d’enchaînement des années a été entièrement réécrite pour supprimer
l’utilisation du fichier recup.tmp. A chaque nouvelle rotation, toutes les variables sont
réinitialisées sauf les variables d’état du sol et les paramètres généraux.
Les routines d’initialisation ont été réorganisées pour faciliter l’enchaînement des années et
les rotations culturales.
• InitialGeneral = initialisation des paramètres généraux
• InitialSol = initialisation des paramètres du sol
• InitialPlante = initialisation d’une plante selon l’espèce
• InitialVariables = initialisation de toutes les variables états journalières
• IniClim = Initialisation des variables climatiques
Integrating the impact of the trees on the crops

La boucle journalière de STICS a été séparée en 2 parties (après KETP) afin d’insérer coté
Hi-sAFe, un module de répartition de l’eau et de l’azote entre la plante et les arbres. Les 2
parties s’appellent Simuler1 et Simuler2.
Simuler1 récupère chaque jour et pour chaque cellule, les résultats de l’influence des arbres,
à savoir :
• L’interception lumineuse
• L’interception de la pluie
• Le stemflow
Simuler1 récupère chaque jour et pour chaque cellule, les résultats du module de partage de
l’eau et de l’azote à savoir :
• Les extractions d’eau de la culture et des arbres
• Les extractions azote de la culture et des arbres

HISAFE STICS
Interception lumière
Début boucle jour Interception pluie
Stemflow
SIMULER1
Répartition de l’eau et de
l’azote entre arbres et Demande en eau et en
culture azote de la culture
Extraction eau et azote

culture et arbres
SIMULER2
Fin boucle jour Teneur en eau et azote du sol

Densité racinaire de la culture
Table 42 : Algorithme simplifié d'enchaînement des modules

entre Hi-sAFe et STICS
La routine LIXIV.C a été modifiée pour la prise en compte de l’extraction d’eau des arbres
dans les transferts d’eau dans les mini-couches de sol.
La boucle journalière a été aussi modifiée pour éviter l’exécution de la routine TRANSPI.c
qui calcule l’extraction réelle de la culture lorsque ces valeurs ont été déjà calculé par Hi-
sAFe (c’est à dire lorsque les arbres extraient aussi de l’eau du sol)
Un module JAVA d’agrégation des valeurs de mini-couches STICS vers les voxels Hi-sAFe a
été écrit afin de répercuter dans Hi-sAFe les résultats de STICS à savoir :
• Les densité de racines de la culture
• Les teneurs en eau
• Les concentration en azote
• Les extractions d’eau et d’azote de la culture (lorsque ces valeurs n’ont pas été
calculées par Hi-sAFe)
Un second module JAVA de désagrégation des valeurs de voxels Hi-sAFe vers les mini-
couches STICS a été écrit afin de répercuter dans STICS les résultats de Hi-sAFe à savoir:
• Les extraction d’eau et d’azote des arbres
• Les extraction d’eau et d’azote de la culture
Ces 2 modules sont exécutés tous les jours et pour chaque cellules influencées par les
arbres.

What’s next ?
• Dans Hi-sAFe les demandes en eau des arbres et de la culture sont réduite par un
facteur dit « de Campbell » qui est peut être redondant avec un calcul déjà effectué
dans STICS. Si la réponse est OUI, voir s’il est possible de le déconnecter.
• Comme pour le bilan d’eau, l’extraction en azote de la plante doit être déconnecté
lorsque le calcul est déjà effectué dans Hi-sAFe. Il faudra pour cela étudier la bilan
azoté de STICS et en particulier la routine ABORSORBN.C
• Le code de STICS doit être corrigé pour rendre actif l’activation de l’option
« remontée capillaire » . L’algorithme devra être modifié pour que les remontées
capillaires s’effectuent à tous les changements d’horizons et non pas seulement au
dernier (à vérifier dans le code de STICS)
• Réincorporation des matières organiques de la sénescence des feuilles et des

racines des arbres dans le SOL
Improving Hi-sAFe to allow runs on several consecutive years
This included some complicated issues about data buffers management. Soil initialisation is
now correct and allows chaining years. Some concerns remain with the tree initialisation, but
should be OK within some weelks.
WP6b: Simple biophysical model Yield-sAFe
Summary
The simple biophysical agroforestry model Yield-sAFe, developed, calibrated and tested by
the WU team, was applied at Landscape Test Sites and experimental network-sites during
the last report period. Since then, the model has been extensively calibrated and applied,
mainly by Anil Graves, Paul Burgess (CRAN), Joao Palma (FAL) and Karel Keesman (WU)
and supported by the partners running experimental sites and/or experts of the Landscape
Test Sites (LTS) in Spain, France and England. The model performance was not only
evaluated by a large consortium of agroforestry experts, but also by the developers of the
complex agroforestry model Hi-sAFe. The contribution of the INRA UMR-SYSTEM (Christian
Dupraz, Martina Mayus, Isabelle Lecomte, Thomas Borell and Fabien Liagre) was larger
than initially planned. Their work included:
• Application/ calibration of Yield-sAFe for Vézénobres
• Evaluation of parameters based on expert knowledge, experimental data and

comparison with Hi-sAFe simulations.
• Evaluation of Yield-sAFe results in terms of biomass production and LER (Land

Equivalent Ratio) statistic data, a comparison with the results of a “LER generator”.
The output of Yield-sAFe is required for the economic analysis of silvoarable scenarios at the
plot (WP7) and farm/ region level (WP8), which in turn forms a crucial information
(environmental and economical impact of agroforestry and the effect that agricultural policy
changes has on these) for Deliverable 9.2, the ‘Agroforestry Policy Options’ document. This
document highly addresses End-Users. Therefore, the performance of Yield-sAFe and Plot-
sAFe has been intensively evaluated by the consortium partners. Related definitions and
underlying concepts were discussed and distinctly defined. A major corner stone for the
silvoarable scenarios is the LER analysis, which was defined differently by the consortium

members. The importance of the definition of the LER became evident during the workshop
in Zurich November 2004. The LER served as a measure for evaluating the model
performances, simply by comparing the simulated LER with the few available literature data
and expertise from network-sites.
During the Zurich workshop, the results of Yield-sAFe and Plot-sAFe for the LTS were
presented for the first time to a large audience. It became evident that the simulated LERs
were often too high. According to the experience of C. Dupraz based on European
silvoarable sites and literature on temperate agroforestry in general, the LER of the LTS and
network scenarios with walnut, poplar, cherry, oak and cereals should vary between 1.0 and
1.4. This hypothesis forms the base of the developed “LER generator”, a tool to generate
sites specific time series of tree and crop yields (Dupraz and Borrel, see report below). The
results of the LER generator for French LTS sites, helped to identify weak points and poor
predictions of Plot-sAFe.
The reason for the high LER was partly due to problems of re-calibrating Yield-sAFe,
calibrated for regional potential yields, to regional realistic yields. So far this was done by
adapting the parameter gamma (water use in m-3 g-1 to produce 1 g of aerial biomass). This
resulted in unrealistic gamma values and yields. Mayus suggested to use a kind of site
factor, that would in fact reduce the light use efficiency (eps) and thus the total aerial
biomass. According to Joao Palma, a set of several parameters should be used for adapting
the model calibrated under potential growth conditions to realistic growth situations. This re-
calibration should be done within a range of reasonable (literature of expert knowledge)
ranges of parameter values. Later realistic boundary values for the values of gamma, eps,
and HI were found in literature and a refined calibration procedure was applied (WP6 report).
The second major problem of the model was the overestimated light interception and initial
growth of trees at low tree densities when using a fixed kt of 0.8. It appeared that a phased kt,
decreasing (e.g. from 0.8 to 0.4) as leaf area increases, showed best results. The need of
changing kt with time was also identified by the application of Yield-sAFe to Vézénobres (see
below). The complex agroforestry model Hi-sAFe was used to deduce values of kt as a
function of tree densities and tree leaf area (see below).
The information of Hi-sAFe with respect to the dependence of kt values on tree density and
tree leaf area was included in a refined calibration procedure. The latter was a joint effort of
Graves, Burgess, Palma and Keesman (December 04) after substantial discussion including
Dupraz, Van Keulen, Van der Werf and Mayus. The new calibrated procedure was tested for
Silsoe and Vézénobres. Finally all LTS scenarios were redone with recalibrated Plot-sAFe
and the results were evaluated by C. Dupraz and F. Liagre and with respect to Vézénobres
also by M. Mayus.
Application of Land Equivalent ratio to agroforestry systems

An agroforestry system consists of a mixture of woody perennial crop and annual crop. For
simplicity the perennial and annual components are called tree and crop, respectively. The
discussion on “how to apply LER for agroforestry” took place between several consortium
partners by means of email. Dupraz, stressed that, “the LER is a global measure of Land
Use Efficiency. It is aiming at answering the management oriented question: Should I mix
trees and crops or not”.
In the specific case of agroforestry, LER is defined as the area of land required by a
monocropping system to produce the same quantity of products as the mixed cropping
system.

The area, means the area has to be fixed when using the LER measure for agroforestry.
Since, in a mixed tree-crop system, the farmer cannot change the area occupied by tree
monocropping every year. The tree monocropping area is determined by the ratio of yields of
the tree component in mixed (or intercropping, I) and in monocropping (or sole cropping S)
(Eq. 1: [treeIyear-n / treeSyear-n]). The yields over a period of n years are the sum of the annual
increments. The equivalent tree monocropping component must therefore have a fixed area.
This is turn imposes a fixed area for the crop in monoculture (or sole cropping S).
Consequently, the LER integrated over n years cannot be computed with the arithmetic
means of the annual LERs, except when the monocropping yields are constant. Christian
Dupraz illustrated this by computing several examples of LER scenarios (SAFE web disk:
The LER with random yields.xls).
A further point that has to be taken into account when computing the LER for agroforestry is,
the final year of intercropping (or switching to a shade tolerant crop as pasture). Farmers
would only continue cropping, when the yields are still economical beneficial. Examples of
threshold yields are: 2 t/ha of durum wheat in Mediterranean France and 4 t/ ha for winter
wheat in England. The threshold will be applied when the yield of the crop is below that
threshold and if this trend is consistent for e.g. two years. In this way it is assured that the
decision to stop intercropping was not related to climate hazards but to the tree
impact.
Consequently the LER for an agroforestry life cycle of n years is computed as:
LER = [treeIyear-n / treeSyear-n] + [(cropIyear1+...+cropIyear-x)/(cropSyear1+...+cropSyear-n)]

(eq. 1)
where
treeI, cropI = yield per ha of trees and crops when intercropped
treeS, cropS = yield per ha of trees and crops when sole cropped
year-x , year-n the final year of intercropping and the agroforestry life cycle, respectively.
Regarding the LER definition above, the products could refer to the total aerial biomass
(Biomass-LER) or to the economic interesting output as timber and grain yield (Main
Products-LER). The Biomass-LER is the better measure for checking the model
performance, since Yield-sAFe simulates the total aerial biomass. Whereas the Main
Products-LER addresses the management question and is required for instance to decide
when to stop intercropping or the year of tree thinning. Thinned tree should be included in the
LER computations too.
Application of Yield SAFE for Vézénobres

The agroforestry system in Vézénobres is the most mature silvoarable system in France/
Europe. This makes it an interesting experiment for the evaluation of Yield-sAFe. The poplar
trees, planted in 1996 and 1997, showed a fast growth and reached a height of about 20 m in
year 8. The model application, to the poplar - durum wheat stand at Vézénobres, located in
Mediterranean France was not easy. The first problem was the weather file, since the
radiation data of Vézénobres covered only several months. Therefore, first simulation tests
were performed with radiation data of the 30 km southwest located experimental site
Restinclières. The simulation was performed for potential growth conditions, i.e. the soil
water balance was not activated. The parameter set calibrated for Mediterranean region was
adapted to the site with respect to management factors as tree density and sowing and
harvest dates. The conversion from aboveground biomass to yield was performed with fixed

relations, i.e. HI for crops and for trees into merchantable volume using ρtimber, the dry wood
density of trees (g m-3) and πtimber, the proportion of the above-ground biomass that forms
timber according to the description by Burgess et al. (2004).
Management settings of the model

Agroforestry life cycle: 12 years, but run for 15 years
Tree density agroforestry: 140 trees/ha.
Tree density pure forest: 210 tees/ha or also 140 trees/ ha
Tree weight at planting was: 100 g/ tree
DOY BudBurst: 95 DOY
DOY Leaffall: 306 –325, measured...
Crop: durum wheat each year
HI: 0.2; computed by STICS
πtimber 0.5
ρtimber 260 kg m-3 (Souleres 1992).
General growth condition

The agroforestry plot was well fertilised and nitrogen stress should be ignorable. Moreover,
diseases and weeds have been chemically controlled and hand weeding took place too in
2004. Nevertheless, in 2003 the weed problem was serious. Poplars have access to water
table, which limits the water stress. This will be later implemented into the model by using a
lower than realistic value for water use of trees, gammat, to mimic the fact that a large part of
water uptake takes place outside the tree-crop root inter-zone.
Results
Table 43: Total aerial biomass and LAI of durum wheat simulated with Yield-sAFe for
potential growth conditions, Vézénobres, Mediterranean France.

Table 44: Total aerial biomass and LAI of durum wheat simulated with Yield-sAFe for
realistic yields, Vézénobres, Mediterranean France.
Monocropping yields
The simulations for potential growth conditions show reasonable estimates for potential
durum wheat yields (Figure 1). As reference for potential yields we took the simulation results
provided by the crop growth model STICS and the expert knowledge of C. Lafon (pers.
comm., 2004), a regional farmer information officer. STICS simulated average potential
yields of 30 t/ha and grain yields of about 8.2 t/ha (with HI ca. 2.7), which is in good
agreement with the Yield-sAFe predictions. The predicted yields for realistic growth
conditions (Figure 2) were with 4.4 t/ha similar to those found in the monoculture wheat crop
in Vézénobres, where the average yields were 4 t/ha with a range of 1 and 5 t/ha. The
realistic yields were achieved by using a site factor, whereas the water module was switched
off. The site factor was a reduction factor multiplied with the light use efficiency, eps. This
was done, because at this stage the water part of the model has not yet been calibrated. At
the Vézénobres site, poplar growth was more or less potential and no further recalibration
was done. The simulated total aerial biomass was with 500 kg/tree a bit overestimated
(Figure 3), the experimental value was about 400 kg/tree. The simulation results per poplar
tree were the same in the agroforestry and in the monocropping stand.

Table 45: Agroforestry yields simulated with Yield-sAFe for realistic growth
conditions, Vézénobres, Mediterranean France. Above total aerial biomass and leaf
area per tree. Below total aerial biomass and LAI of durum wheat.
Agroforestry or mixed cropping yields

The crop yield in the alley of poplar trees was hardly affected during the first four years of the
agroforestry system (i.e. 1997 –2000). Subsequently, with the increase of the poplar canopy
size the grain yield was progressively reduced. In 2001, the reduction was about 20% (3.7
t/ha AFS and 5 t/ha crop control), in 2003 and 2004 about 50%. In 2003, the crop yield was
2.1 and 4.1 when intercropped and sole cropped, respectively. In 2004 the yields were 2 t/ha
and 4 t/ha intercropped and sole cropped. The simulated competition effect of trees was
underestimated. Apparently, it is not possible to achieve good Yield-sAFe predictions for the
poplar-wheat site at Vézénobres when using the parameter set for Mediterranean region. A
small decrease after 4 years could be achieved only by recalibration, the parameters that
have an impact of light interception (Figure 3). In particular the parameter, kt, which highly
determines the amount of light available for the intercrop has an impact (see also sensitivity
analysis). The simulation run tests clearly indicated that the parameter should be different for
young and tall trees. Accordingly other parameters might change, like the one determining
light use efficiency.
In the meantime, the model has been extended to allow phased kt and simulations including
water stress, but the simulation runs were not yet performed, since we waited for the new
calibration procedure recently finished by the team of Graves, Burgess, Palma and
Keesman. The simulations and its analyse are in work for Vézénobres and Restinclières.

Using the Hi-sAFe model to improve the simplified Yield-sAFe above-
ground model by calibrating the kt parameter of Yield-sAFe
kt is the radiation extinction coefficient of the tree leaf canopy and it appears in the Yield-
sAFe equation that predicts the fraction of radiation intercepted by the trees in the
−k L
agroforestry system : f t = 1 − e t t where Lt is the leaf area index of the tree stand (m2 tree
leaf area per m2 silvoarable stand). The Yield-sAFe model does not include an explicit
desegregation of the tree canopies over the silvoarable plot, and assumes that the tree leaf
area is spread over the whole agroforestry plot. The assumption made in Yield-sAFe was
that the usual extinction coefficients of tree canopies could be applied to this equivalent leaf
area over the whole silvoarable scene.
The kt parameters has a large relative effect on the predicted LER (cf. WP6b report) . The
nominal value assumed was 0.8. However, this resulted in a questionable pattern of tree-
crop competition: it resulted in a long term overestimation of tree growth. Reasoning from
existing literature on light distribution in crops indicates that the extinction coefficient might
change at low tree densities as the canopy is more heterogeneous. Initial use of the model
has suggested that it may be necessary to modify the light extinction coefficient in such
situations.
The Hi-sAFe model was therefore used to calculate the Yield-sAFe kt value for a range of
tree densities (10 – 50 – 100 – 400 trees per hectare) and of tree sizes (1 – 10 – 100 – 300
m2 of leaf area per tree). We used the Hi-sAFe model to calculate the average radiation
available on the ground of the silvoarable plot at the day time step. Then a kt value was fitted
to the Yield-sAFe equation so that the distance between the two estimates is minimized over
the whole growing season of the trees. The winter period with no tree leaves was not
included in the fitting calculation. Hi-sAFe settings included a fixed leaf interception factor of
0.85, tree canopies were assumed to be ellipsoids with an homogeneous leaf area density,
the diffuse radiation was modelled by a turtle with 48 directions and the Standard OverCast
(SOC) distribution, and the direct radiation was recalculated when the tree leaf area had
changed by 5% or the sun elevation had changed by 2 degrees.
1,00 16
0,96
Tree Leaf area (m²)
Light transmission
12 Hi-sAFe
0,92
Yield-sAFe kt = 0.54
8
0,88 Yield-sAFe kt = 0.8
4 Tree leaf area
0,84
0,80 0
1 61 121 181 241 301 361
DOY
100
trees/ha of 14 m² leaf area per tree resulted in a fitted kt = 0.54
Table 46 : Fitting a kt value to match Plot-sAFe predictions and Hi-sAFe predictions of

radiation availability at the ground level in a silvoarable plot
Adequate Hi-sAFe settings were used to "see" small trees in a huge scene for low density /
small trees scenes. There is a limit in Hi-sAFe for the LAI of trees, due to tree-tree
competition : we could not get tree LAI values above 6,6. This means that the "1000 m²"

trees are in fact smaller at density of 100 and above. The leaf area of a single tree is
indicated in the last table.
Hi-sAFe is taking into account the 3D geometry of the system. A clear example of this is that
when the leaf area of the tree is constant (after the end of the long shoots expansion until the
beginning of the leaf fall), the leaf interception of the tree varies with the sun declination.
Yield-sAFe cannot take this into account, and predicts a stable light interception when the
leaf area is stable, irrespective of the sun elevation.
Fitted kt Tree leaf area (m²)

1 10 100 300 1000
Tree 10 0,31 0,26 0,18
density 100 0,54 0,31 0,29
(tree ha- 400 0,85 0,58 0,48
1
)
Cells in red could not be simulated with Hi-sAFe due to the very small size of the trees; pink cells correspond to
implausible tree stands (too high density for the size of individual trees)
Table 47 : kt values for Yield-sAFe predicted for various silvoarable stands of walnut
trees
It finally appears that appropriate kt values for the Yield-sAFe model should not be constant
with the age of the tree. The suitable range was from 0.85 for very small trees to about 0.2
for very large trees.
As a consequence, using Yield-sAFe with a fixed value of kt across the tree life induced an
overestimation of the interception of the tree light by the tree component when the trees
become large, and this was observed in the former Yield-sAFe calculations. Using the
nominal and constant value of 0.8 resulted in large overestimations of the interception of light
by mature trees (Table 46).
Similarly, using Yield-sAFe with a fixed value of kt across the growing season induces an
underestimation of the tree light interception in spring and autumn (low sun elevation), and
an overestimation in summer (high sun elevation). This may result in Yield-sAFe unduly
favouring winter crops.
The impact of the tree density on the optimised value for kt was less noticeable. kt is slightly
increasing with tree density for medium size trees.
If we ignore the seasonal (sun elevation) effect on kt and look for a stable annual value that
would depend on the tree size and density, the following equation was produced:
kt =-0,187 Log(Tree Leaf Area)+0,000299 Tree density+0,717 with r²=0,97

1,00
Kt predicted from Tree leaf area and

0,75
tree density
0,50
y = 0,9668x + 0,014
0,25
R2 = 0,9668
0,00
0,00 0,25 0,50 0,75 1,00
Kt deduced from Hi-SAFE runs
Table 48: Comparison of predicted kt values from tree density and tree leaf area with
Hi-sAFe calculated values.
This suggestion of a phased kt that decreases with time was finally included in Yield-sAFe.
1,0 0
100 trees/ha
0 ,8 0 10 trees/ha
400 trees/ha
0 ,6 0
0 ,4 0
0 ,2 0
0 ,0 0
0 ,0 0 0 ,5 0 1,0 0 1,5 0 2 ,0 0 2 ,5 0 3 ,0 0 3 ,5 0
Log(Tree leaf area)
Table 49 : Fitted kt are well correlated with the leaf area of trees.
Discussion
This was the first cooperation between the two biophysical SAFE models Hi-sAFe and Yield-
sAFe. It was performed by Christian Dupraz and Isabelle Lecomte in consultation with Paul
Burgess.
A declining value of kt with increasing leaf area appears to result in more satisfactorily tree
growth pattern (in particular at low tree stands) and corresponding intercrop yields than using
a fixed value (Burgess and Graves, 2004). The reason for this is self-shading: in large tree
canopies not all leaves are exposed to full sunlight, hence, not all leaves contribute much to
the light interception. A phased kt would mimic the effect of the declining "canopy cover"
value (C; values ranging from 0 to 1) found in the Jackson and Palmer model of the

proportion of light (fs) intercepted with wider-spaced trees (fs = C (1 - EXP (-k LAI)).
(Burgess, email 11 Nov04).
The problem of a fixed light extinction coefficient in tree growth modelling (SUCROS type)
was also experienced by Cittadini (2002). He applied the theoretical kt value of 0.7 for
spherical leaf angle distribution (Goudriaan and van Laar, 1994) for a sweet cherry orchard
(3.25 m between rows and 0.6, 0.9, 1.2 and 1.5 m between trees in the row). Because of the
row arrangement a clustering factor CLF was calculated with a sub model (Goudriaan pers;
comm.) and multiplied by k. The CLF was 0.95, thus the k’ = CLF * k = 0.665. Using this k’
the simulated radiation interception underestimated the measured ones in the beginning of
the growing season and overestimated the interception around fruit harvest.
Accurate values of kt are difficult to obtain, and only few data can be found in literature.
Values reported for broad-leaved tree species ranged roughly from 0.4 to 0.7 (Jarvis and
Leverenz, 1983, Gazarini et al. 1990, Vose et al., 1995), but information on how these values
depend on tree density and LAI could not been found. Since no literature or experimental
data were available the complex Hi-sAFe model was applied to deduce the parameter value
as a function of tree density and LAI.
Moreover, for a wide spaced tree stand with a given tree leaf area, kt appears to vary with the
sun elevation. In case of low sun elevation kt should be high and vice versa. This suggests
that the model should use different kt for winter and summer periods, otherwise the available
radiation for the crop would be overestimated in winter and underestimated in summer.
Consequently, in the overall system analysis of Yield-sAFe and FARM-SAFE, wintercrops
would be favoured in terms of biomass production. This aspect has not yet been further
investigated. It must be noted, that trees in Yield-sAFe intercept only during the leafy tree
period, which is roughly between April and October (DOY 100 and 300). The effect of sun
elevation for various latitudes should be considered in the Yield-sAFe calibration and
eventual tested with further Hi-sAFe runs. In Hi-sAFe, which considers the 3D geometry of
the system, this problem does not occur. This was tested at the example of a tree having a
constant leaf area (after the end of the long shoots expansion until the beginning of the leaf
fall); the leaf interception of the tree varies with the sun declination, as it should.
We get different kt values for a poplar, even if we keep with the same leaf scale kt of 0.85 in
Hi-sAFe. The poplar canopy has a very different shape, and for the same leaf area would be
represented by a very different ellipsoid. It can concluded, that the parameters for the kt
variation are species-specific. Further results and information on the method can be found on
the safe WEB SITE (private disk space).
Sensitivity analysis: Task 6.6

The model evaluation includes a sensitivity analysis (SA), which is described in detail in the
third annual report of SAFE and in the paper: “Yield-sAFe, A parameter-sparse process-
based model for calculating growth, yield and resource use in agroforestry systems (Van der
Werf et al. in progress). The simulation analyses refer to a poplar - winter wheat system
under optimum growth conditions of the Leeds experimental site. Due to the low tree density
(156 trees/ha), tree to tree competition does not occur. During the course of the model
application it became evident that the tree factor responsible for light interception (kt) was
sensible and rather uncertain. Two major values appeared to be important: a kt value of 0.4
and a kt value of 0.8. In the last report period the SA was performed for a kt value of 0.8 and
in this report period for a kt value of 0.4 (Table 2 and 3). For the latter the tree part was
recalibrated for the same data set, resulting in other values for εt and N[t0] (Table 1). Overall,
the results are similar to the one presented in the last report, namely: The tree factors have
the largest influence on the land equivalent ratio (LER) and factors influencing the light
interception and light use are in this simple model of major importance (Table 3). However, in
case of scenario kt = 0.8, the crop parameters kt and εt play play a large role in year 20 and

25 compared to the other scenario. The reason is that tree light interception is double of that
with kt of 0.4. During the entire period of 25 years, higher poplar and wheat yields as well as
a higher LER are predicted with simulation scenario kt =0.4 than with scenario kt =0.8.
Table 50: Values for parameters and initial conditions for the sensitivity analysis of
Yield-sAFe for a poplar agroforestry stand with continuous wheat after calibration for
the site in Leeds, UK.
Symbol Unit Description Default Default

value value
TREE scenario scenario
kt = 0.4 kt = 0.8
εt g/ MJ Radiation use efficiency tree 1.84 1.092
kt - Light extinction coefficient tree 0.4 0.8
Am M2/ shoot Maximum leaf area of single 0.05 0.05
shoot
τ D Time constant of leaf area 10 10
growth of a tree shoot
α d-1 Maintenance respiration 0 0
coefficient
N[t0] # / tree Initial number of shoots per 1.32 1.075
tree
Bt[t0] g/ tree Initial Tree biomass 100 100
Lt[t0] M2/ tree Initial LA per tree 0 0
Nm # / tree Maximum number of shoots 8000 10000
per tree
tBudBurst DOY DOY of budburst 100 100
tLeafFall DOY DOY of leaffall 265 265
CROP
εc g/ MJ Radiation use efficiency crop 1.6 1.6
kc - Light extinction coefficient crop 0.7 0.7
σ m2/ g Specific leaf area crop 0.02 0.02
P - Initial Biomass partitioning 0.8 0.8
factor to leaves
o
T0 C Lower threshold for crop 0 0
phenological development
o
Sh Cd Heat sum at crop harvest 2950 2950
o
S1 Cd Heat sum at which partitioning 160 160
to leaves starts to decrease
o
S2 Cd Heat sum at which partitioning 2350 2350
to leaves starts to ceases
o
S2 Cd Heat sum at which crop 150 150
emerged
Semergence
Lc[t0] - Initial crop leaf area 0.1 0.1
Bc[t0] g/ m2 Initial crop biomass 10 10

Table 51: Ranking of elasticities of land equivalent ratio’s in years 2, 5, 10, 20 and 25
of a poplar-wheat agroforestry stand to biological parameters of tree and crop for
scenario kt = 0.4. Ranking over all parameters per year: Highest rank 1 to 4; then
rank 5-8, rank 9-12.
Parameter Rank of elasticity, eLER

Type year 2 year 5 year 10 year 20 year 25
TREE
Bt[t0] 7 6 7 9 11
kt 3 4 3 4 2
εt 8 7 6 8 10
N[t0] 5 5 5 7 9
Am 4 3 4 3 3
τ 9 9 10 11 12
Nm 17 12 9 6 4
tBudBurst 1 1 1 1 1
tLeafFall 6 2 2 5 5
CROP
BC[t0] 14 15 13 13 15
LC[t0] 15 16 15 15 16
kc 10 10 11 10 6
εc 11 11 12 12 8
σ 16 17 17 17 17
S1 12 13 16 16 13
S2 13 14 14 14 14
P 2 8 8 2 7
Future Actions include:
1. Evaluation of the Yield-sAFe description of light interception by trees by comparison

with the Hi-sAFe light model.
2. Comparison of Hi-sAFe and Yield-sAFe – Field scale for wheat-poplar in NE (UK) and
wheat-walnut SE (France).
3. Finish writing of scientific papers on modelling and calibration.
Reference
Burgess, P. and Graves, A. 2004. Draft Discussion on the value for the extinction coefficient
(kt) in Yield-sAFe.
Cittadini, E. D., 2002. Development of a mechanistic simulation model for potential

production of sweet cherry: its usefulness to analyse planting density. MSc thesis, Group
Production Ecology Wageningen University, The Netherlands.
Gazarini, L . C ., Araujo, C.C., Borralho, N. and Pereira J.S. (1990). Plant area index in
Eucalyptus globulus plantations determined indirectly by a light interception method. Tree
Physiology, 7:107-113.

Vose, J. M., Sullivan, N.H., Clinton, B.D. and Bolstad, P.V., 1995. Vertical leaf area
distribution, light transmittance, and application of the Beer-Lambert law in four mature
hardwoods stands in southern Appalachians. Can. J. For. Res. 25: 1036-1043.
WP7. Economics of silvoarable agroforestry : Using a priori LERs

INRA in cooperation with APCA explored an alternate way to produce time-series of tree and
crop yields in agroforestry systems. This approach was decided when the prospect for
obtaining satisfactory biophysical models was quite dim, at the Toulouse modelling workshop
in April 2004. The concept was then considered worth exploring. The basic idea is as follows:
If we know the integrated LER of a silvoarable system, it becomes possible to generate time
series of tree and crop yields over time, using the LER as a forcing value. This approach
does not predict LERs, but generate yield series that match a given LER. With the yield
series, economics studies can then be conducted with Farm-sAFe, allowing performing a
sensitivity analysis of the profitability of the system to the biophysical efficiency of the
system.
Improving the LER concept for economic studies
LER-biomass and LER-product

The Land Equivalent Ratio indicates the area of monocultures needed to produce as much
as one intercropped hectare (Vandermeer, 1989). It is calculated as the sum of relative areas
(RA), i.e. productions ratios: for each product, the intercrop production divided by the
monoculture production. In most of the agroforestry cases, there are 2 RAs: the crop RA and
the tree RA. For instance, a tree RA of 0.7 means that an agroforestry plot produces as
much timber as a forestry plot of 0,7 ha. A LER of 1.3 thus indicates than intercropping
produces 30% more than monocropping.
However, it can be calculated either with total biomass or only with commercial products,
particularly in the case of timber production: the higher rate of thinning in forestry than in
agroforestry implies different tree relative yields whether it is calculated with or without
thinned trees.
This distinction leads to two different indicators: the LER-products, calculated with the
commercial products (bole of timber of the felled trees, grain of the cereals, etc.), and the
LER-biomass, calculated with the total biomass produced on the plot (for their detailed way
of calculation, see Dupraz et al., 2005). Although the likely range of values for the LER-
products is still to be defined with experimental plots and models, we already know that the
expected values of LER-biomass are likely to be comprised between 1 and 1.4. Indeed, a
value below 1 is biologically unrealistic considering that if one of the intercrops dominates too
much the other, it shall perform as in a monoculture plot and thus produce as much biomass
of the same area of monoculture production. A value above 1.4 seems too much optimistic
with regards to present experimental results and bibliographical documentation (Dupraz et
al., 2005).
The LER-based generator

For this study, we used a constrained generator of data: forest, arable and agroforest time-
series are generated in accordance with an expected LER-biomass (see Annex 1: Detailed
description of the LERbased-Generator).
The tree RA-biomass is defined according to the densities in forestry and agroforestry and to
the expected increase in tree growth rate at low density. The crop RA-biomass is then
deduced in order to reach the predetermined LER-biomass. The LER is only divided in a
crop component and a tree component (timber); it is thus impossible to generate data sets

for a third component (fruits for example), such as for a traditional orchard or double purpose
walnuts.
The arable and forest reference data and the values of these two RAs permit to generate all
the time-series under constraint:
o the arable time-series is the repetition of the reference yields in accordance with the
rotation;
o the forest time-series is generated in function of the reference volume of timber per
ha at felling;
o the two agroforest times-series (one for the intercrop, one for the trees) are generated
so that the constraint fixed by the RAs is respected (sum of productions for the
intercrop, volume of timber per ha at felling for the trees).
Amongst the hypothesis made in this generator, we assume that :
o the agroforest trees are felled at the same time as the forest trees, but their higher
growth induces bigger individual pieces of timber; In any case, the unit volume in
agroforestry doesn’t exceed 20 % of the forestry volume one.
o there is no difference in the partition of biomass for the intercrop and a classical
arable crop: the crop RA-products is thus equal to the crop RA-biomass, which shall
both be called “crop RA”;
o the intercrops cannot offer higher yields than the arable crops without any tree
(consequently the value of the crop RA cannot be superior to the maximum
intercropping area: 1 – the proportion of area occupied by the tree strips); we made
therefore the hypothesis that the trees don’t affect positively the crop yield which
could be discuss on a long term period (soil erosion and fertility, wind effect, etc.).
o the width of the intercropped alley can be reduced by successive steps when the
yield decreases (less productive areas are given up), in order to preserve
economically acceptable yields as long as possible. When it cannot be reduced
anymore (at a minimum width), the intercrop is suppressed when it is no more
profitable (profitability threshold yield).
A wise hypothesis for the agroforest tree growth

In most of the cases, agroforest designs are at lower density than forest designs. As a
consequence, trees grow quicker. We assume that the growth rate increases when the
density decreases, until to reach a critical final density where the genetic potential is fully
expressed. Below this density, we assume that trees don’t grow more, even if they are
completely isolated.
At this critical density, we assume that trees grow at a rate driven by a coefficient: the
individual tree timber volume growth acceleration in low density AF conditions, or Tree
Growth Acceleration (TGA). At critical density, the volume of an agroforest individual piece of
timber at felling, VAF, is thus calculated as:
VmaxAF =TGA x VF
where VF = volume of a forest individual piece of timber at felling

(in the forestry reference data which is used).

VmaxAF is thus the maximum volume of an individual piece of timber.
Unfortunately, the critical densities and the likely range of values for TGA are not well
documented. Thus these parameters had to be fixed by expert knowledge.
In order to realise wise simulations, we assumed a quite low value of TGA: 1,2 for the three
species (Table LII).
Species Final density in forestry VF TGA Critical density VmaxAF

Walnut 100 trees/ha 1 m3/tree 1,2 50 trees/ha 1,2 m3/tree
Wild cherry 150 trees/ha 0,8 1,2 60 trees/ha 0,96 m3/tree
m3/tree
Poplar 200 trees/ha 1,5 1,2 100 trees/ha 1,8 m3/tree
m3/tree
VF is the volume of timber of an individual forest tree. TGA is individual Tree timber volume Growth
Acceleration in agroforestry at densities lower or equal to the critical density. The critical density is the highest
density at which the maximum volume of an individual piece of timber is reached. VmaxAF is the maximum timber
volume of an agroforest tree, reached at densities lower or equal to the critical density.
Table LII: reference values in forestry and values of TGA, the critical density and
VmaxAF for each of three tree species
Nevertheless, some unpublished experimental results are in favour of higher values for TGA:
at M. Jollet’s farm (Les Eduts, Charentes Maritimes, France), INRA’s measurements of the
forest and agroforest trees at the middle of the revolution indicate a TGA of 2 for black
walnuts, at 80 trees/ha (Gavaland, pers. com.). But another thinning will soon accelerate the
growth of the forest trees, and then this estimated TGA is likely to decrease.
There is thus an important difference between our hypothesis and what we could expect
(Table 53).
individual piece of timber (m3/ tree)
2 with TGA = 2
with TGA = 1,2
1,5
0,5
0 20 40 60 80 100
Final density (trees/ha)
Table 53: Volume of the individual walnut timber volume in function of the final density
and of the value of TGA.
As an economic consequence of such a wise hypothesis, the volume of timber at felling is
less important, thus the revenue of the tree component might be under-estimated.

Maximum expectable LER in function of species and final density
As the production of agroforest timber is determined in accordance with the densities, the
critical density and TGA, the tree RA-products and the tree RA-biomass are fixed: it is
impossible to tune them without modifying one of these previous parameters. Then the range
of variation of the LER (biomass or products) corresponds to the crop RA:
• As a LER-biomass inferior to 1 is biologically unrealistic, the minimum value of the

crop RA is equal to 1 – tree RA-biomass.
• As we assume lower or equal yields, the maximum value of the crop RA must be
inferior or equal to the maximum intercropping area. In our optimistic assumptions, at
highest densities (tree lines every 10 m), we assumed a crop RY at ¾ of the
maximum intercropping area.
• A likely value would be the mean of these two extreme values.
As the proportion of land required by the trees strips rises with the density, the maximum
crop RA decreases when the density gets higher. A first conclusion is that we obtain
acceptable RA with densities, which correspond to distances between the tree lines included
between 24 to 40 m.
1,1
0,9
0,8
0,7 LER max

0,6 LER-optimist
LER-medium
0,5
LER-pessimist
0,4
0,3
0,2
No tree
0 20 40 60 80 100 120 140
Distance between the trees lines (m)
Table 54: Range of values of the crop RA with walnut, wild cherry and poplar, according the
distance of the tree lines and depending on how optimistic the dynamic of the LER is.
In forestry, the realisation of many thinnings means that a lot of biomass is synthesised in
addition of the trees, which shall be conserved until the last fall. As we assume that the
volume of the agroforest trees is maximum 20% bigger than the one of the forest trees, the
production of woody biomass is small compared to the one of a forestry plot. Thus the ratio
of woody biomass, i.e. the tree RA-biomass, is low. At low density, even an optimistic value
of the crop RA is insufficient to compensate such a low tree RA. Consequently, high LER-
biomass cannot be reached for all densities, in particular for species with a high rate of
thinning in forestry such as wild cherry (Table 55).
However, very satisfactory LER-products can be reached even with these species.

1,6
1,4
LER - Products
1,2
Optimist Crop RA
1 Medium Crop RA
Pessimist Crop RA
0,8
0,6
0 20 40 60 80 100 120 140
Tree final density (trees/ha)
Table 55: Expectable LER-products for walnut, cherry and poplar, depending on how optimistic
the dynamic of the intercrop is
Impact of the TGA on the LER results

In our simulations, we used a TGA of 1.20. We were cautious in our predictions if we
consider some experimental plots (such in Restinclières in France) or private site (Farm of
Claude Jollet in Charente Maritime) where we observed some TGA which reach 2. If we had
taken this value of 2, the tree RA would have increased between 15 to 30 % in comparison
with what we obtained with 1.20.
0,9
y = 0,4191x + 0,1123
R2 = 0,9995
0,8
120 trees/ha - products

0,7
y = 0,2745x + 0,0955
2
R = 0,9987 120 trees/ha - biomass
0,6
TREE RA
0,5 50 trees/ha - products

y = 0,2655x + 0,0018
R2 = 0,9989
0,4 50 trees/ha - biomass
y = 0,1773x + 0,005
0,3 2
R = 0,9965
0,2
0,1
0
0,8 1 1,2 1,4 1,6 1,8 2 2,2
Tree Growth Acceleration
Table 56: Influence of the Tree Growth Acceleration on the tree RA (biomass and
products), according to the tree density.

Data references and main hypothesis
The forestry references

The revolution duration, timber production and production techniques (initial density,
prunings, thinnings, sward maintenance, and final density) were determined by expert
knowledge, in accordance with available documentation.
The densities correspond to the schedule of conditions of the French circular “Forêts de
production” and to forestry organisms’ advises.
Individual density Revolution duration Mean annual production

piece of (trees/ha) Volume (years) (m3/ha/year)
timber at at felling Good Land Bad Good Land
Bad
felling initial final (m3/ha) land unit land land unit
land unit
(m3/tree) unit medium unit unit medium
Walnut 1 200 100 100 46 53 60 2,17 1,89 1,67
Wild cherry 0,8 800 150 120 50 55 60 2,40 2,18 2
Poplar 1,5 200 200 300 19 22 25 15,79 13,64 12
Table 57 : Densities, revolution duration and mean annual and total productions for
walnut, wild cherry and poplar. With walnut, 2 thinnings of 50 trees/ha are realised at
1/3 and 2/3 of the revolution; with wild cherry, 3 thinnings are realised at 1/3 (400
trees/ha), half (200 trees/ha) and 2/3 (50 trees/ha) of the revolution.
Supports for afforestation on agricultural land vary in function of the region and of the tree
species: as the poplar revolution is shorter, the Compensation Payment for Agricultural Loss
(PCPR) is available for 7 years instead of 10.
Type of farm Ly-Lc Hy-Hc

Hy-Lc
(region) (Poitou- (Franche-
(Centre)
Charentes) Comté)
Walnut and wild cherry
Establishment grant (4 first years) 50% of the costs 50% of the costs 0
PCPR farmer (10 first years) 240 €/ha 300 €/ha 0
Poplar
Table 58: Regional supports for afforestation on agricultural land for walnut, wild
cherry and poplar (year 2003). The PCPR is the Compensation Payment for the loss of
agricultural income.
Franche-Comté is a particular region. More than 50% of the area are already woodlands,
thus afforestation is not encouraged: there is no support available for new forestry plantation.
Everywhere in France, newly afforested plots benefit from an exemption from land tax: for
10 years with poplar, 50 years with walnut and wild cherry. In our simulations, this land tax is
comprised between 30 €/ha (Centre) and 39 €/ha (Poitou-Charentes).
Reference data in agriculture

All arable data come from the Farm observatory ROSACE, a tool produced by APCA.
Thanks to this typology of farms made by the regional Chamber of Agriculture, several types
of farms are defined and described, each one corresponding to the mean of 5 to 10 farms
selected by the Chambers experts. Each year, the economical inputs are re-calculated (yield,
net margin, farm costs, labour and CAP payment). In addition, all the technical orientations
and strategies of the farm are also described.
We selected 3 types of farm, which we shall now designate with 4 initials:

• Hy-Lc: High yields and Low fixed costs
• Hy-Hc: High yields but High fixed costs
• Ly-Lc: Low yields and Low fixed costs
For each of them, the ROSACE typology indicates:
• The cropping area of the farm, distinguishing tenant farming and property;
• The crop rotation in function of the quality of the soil (up to 3 Land Units: best,
medium, worst);
• The mean yields, attributed to the medium Land Unit (for the best and worst Land
Units, we respectively assumed an increase and a decrease of 10% of the mean
yields);
• The variable costs, assignable fixed costs and fixed costs and labour.
• The prices of the products and sub-products (straws of the wheat) and the CAP
payments of the farm Single Farm Payment, SFP).
To elaborate the selection of each type of farms, various partners from the Chambers of
Agriculture have participated: Camille Laborie, who is in charge of ROSACE in APCA, Anne-
Marie Meudre (Franche Comté), Catherine Micheluzzi (Poitou-Charentes) and Benoît Tassin
(Centre).

Cropping area of
the farm (ha)
Gross Total
Fixed Mean Net
Typical Margi area Net
costs Crop yield Marg
rotation(s) n (ha) Margin
property
(€/farm) (t/ha) (€/ha)
farming
tenant (€/ha) (€/farm)
Total
(a) wheat 8
wheat wheat (straw 983 776 42,0
oilseed 2 t/ha)
72 8 80 36805 18045
or oilseed 4 918 711 30,0
Hy-Lc
(b) wheat set

oilseed – 323 116 8,0
aside
6,5
wheat 2 t/ha)
sunflower oilseed 3,2 674 422 14,1
56,4 37,6 94 28785 17153
wheat sunflow
oilseed 2,5 799 547 28,2
er
Ly-Lc
sunflower set
– 318 66 9,4
aside
(a) wheat 6,7
wheat 2 t/ha)
wheat oilseed 3,5 728 413 14,1
wheat maize 7,5 555 240 28,2
97,5 32,5 130 40370 12031
maize
or
set
(b) wheat – 313 -2 9,4
Hy-Hc
aside
wheat
oilseed
Table 59: Main economic data and total net margin (€/farm) for every type of farm.
Rotation (a) corresponds to the best land units, rotation (b) to the worst. Set aside is
realised on 10% of the total farm area.
The Net Margin is equal to the Gross Margin minus the fixed assignable costs (land tax and
machinery costs). The Total Net Margin is equal to the Net Margin minus the fixed costs (rent
of land, amortisation and maintenance of the buildings, social contributions, banking costs).
Labour costs are not taken into account.
The profitability threshold yield

With the development of the trees, the crop yield decreases progressively. Below a certain
level, the crop is not more profitable, above al near the tree area. For each crop of the three
types of farm, the threshold yield was first determined according to the price of the product,
the CAP payment and the variable costs, assignable fixed costs and a part of the fixed
costs1. As the results, in proportion of the mean yield of each crop, were roughly the same in
the three farms, we fixed this proportion in order to facilitate the extrapolation to other types
of farm.
1
If the crop is abandoned on a part of the cropping area, we assume that the fixed costs should
decrease a little; thus they must be taken into account in the calculation of this threshold yield.

Mean yield Profitability Mean yield in Profitability threshold
crop
in the farm threshold yield Hy-Hc yield in Hy-Hc
Winter wheat 100 % 50 % 6,7 t/ha 3,35 t/ha
Maize 100 % 70 % 7,5 t/ha 5,25 t/ha
Oilseed rape 100 % 60 % 3,5 t/ha 2,1 t/ha
Table 60: Profitability threshold yield in proportion of the mean yield in the farm and
example for the farm Hy-Hc
The threshold yield is the same in every farm, whatever the land unit is. Thus it shall be
reached more quickly in the worst land unit than in the best land unit.
Main management features of the agroforestry systems

For each type of farm, we simulated the introduction of 2 agroforestry designs in the 3 land
units (best, medium, and worst):
• Plantation at 50 trees/ha, on 40 m spaced tree-lines;
• Plantation at 120 trees/ha, on 22 m spaced tree-lines.
The tree strip is 2 m wide. The width of the intercropped alley is respectively of 38 m and 20
m, thus the maximum crop area represents 95% of the initial area at 50 trees/ha and 91% at
113 trees/ha.
With walnut and wild cherry, an early thinning is realised when the timber volume reaches
0,1 m3 (around the years 10-13), therefore the final densities are different from the poplars’
one (see Table 5).
Agroforestry Forestry
Tree
Density Timber Density Timber Tree RA-
Productio Productio RA-
(trees/ha) volume (trees/ha) volume products
(m3/ha) (m3/ha) biomass
initial final (m3/tree) initial final (m3/tree)
50 40 1,2 48 0,36 0,48
Walnut 200 100 1 100
120 80 1,08 86 0,66 0,86
Wild 50 40 0,96 38 0,22 0,32
800 150 0,8 120
cherry 120 80 0,92 74 0,42 0,62
50 50 1,8 90 0,3 0,3
Poplar 200 200 1,5 300
120 113 1,76 199 0,66 0,66
Table 61: Initial and final densities, volume of an individual piece of timber and
production in forestry and in the simulated agroforestry systems; tree Relative Area
(RA)-biomass and tree RA-products
The crops Relative Areas (RA) have been fixed for 3 hypotheses: optimistic, probable and
pessimistic.
The pessimistic hypothesis means that the LER-biomass is equal to 1. Therefore, the crop
RA is equal to: (1 – tree RA-biomass).
The optimistic crop RA is determined according to 2 constraints:
• The crop RA must be inferior to the maximum intercropping area
• We also assumed to fix a ceiling for the LER-biomass of 1.4. Thus the crop RA is
equal to: (1.4 – tree RA-biomass). This ceiling of 1.4 was reached with walnut and

poplar at 120 trees/ha, so the crop RA seems quite low with regards to the maximum
intercropping area.
We assumed a probable crop RA as the arithmetic average of the 2 previous values

(pessimistic and optimistic) (see Table 6 and Table 7).
Width
Initial Width of the Maximum
between Pessimistic Probable Optimistic
density intercropped intercropping
tree lines crop RA crop RA crop RA
(trees/ha) alley (m) area
(m)
50 40 38 0,95 0,64 0,79 0,94
Walnut
120 22 20 0,91 0,34 0,54 0,74
Wild 50 40 38 0,95 0,78 0,85 0,93
cherry 120 22 20 0,91 0,57 0,72 0,87
50 40 38 0,95 0,7 0,8 0,9
Poplar
120 22 20 0,91 0,34 0,54 0,74
Table 62: Crop RA in function of the tree species, density and optimism level. Bold
values are those, which depend on the ceiling of 1.4 for the LER-biomass.
Width Width of LER-biomass reached with LER-products reached with

Initial between the the the
density tree intercrop
(trees/ha) lines Pessim. probabl Optimist Pessim. probabl Optimist
ped alley
crop RA crop RA crop RA crop RA crop RA crop RA
(m) (m)
Walnut 50 40 38 1 1,15 1,3 1,12 1,27 1,42
120 22 20 1 1,2 1,4 1,2 1,4 1,6
Wild 50 40 38 1 1,07 1,15 1,1 1,17 1,25
cherry 120 22 20 1 1,15 1,3 1,19 1,34 1,49
Poplar 50 40 38 1 1,1 1,2 1 1,1 1,2
120 22 20 1 1,2 1,4 1 1,2 1,4
Table 63: LER-biomass and LER-products in function of the tree species, density and
hypothesis of optimism for the intercrop bold values are those, which depend on the
ceiling of 1.4 for the LER-biomass.
Economic hypothesis
CAP payments
In agriculture, the crops area benefits from the Single Farm Payment (SFP): it was calculated
on the basis of the historical references of each farm, in accordance with the way France
decided to implement the new CAP in 2006.
In the basic scenario, we assumed that the intercrops are eligible to the SFP proportionally to
the area of the plot that they occupy. It is the present situation in France. The rights
corresponding to the tree area could be transferable to another eligible area, which doesn’t
benefit from a payment right. In our simulations, we did not attribute them to new plots,
considering therefore that these rights were lost for the farmer.
Tree grants
In our basic scenario, agroforest trees benefit from the same establishment payments as the
forest trees: 50% of the costs of the 4 first years in Poitou-Charentes and Centre. It
corresponds to the present situation, permitted by the circular “Forêts de protection” which
relies on the line i of the French National Rural Development Programme. However an
agroforest plot can benefit from neither the PCPR nor the exemption of land tax.

In France, an agro-environmental measure called “agroforest habitats” can be contracted
under certain conditions, but it still faces administrative difficulties and is not available in most
of the departments, thus it was not taken into account in our simulations.
Costs and prices

Some key points have to be underlined:
• The cost of sward maintenance is higher in forestry than in agroforestry. In forestry, at

the beginning of the revolution, sward maintenance is realised thanks to two grindings
instead of one for the maintenance of the tree strip in agroforestry.
• The farmer makes all operations himself, except the marking out and plantation of the
young trees. Both of these operations are charged 15 €/h. The timber prices
correspond to standing trees, thus neither the harvesting cost is taken into account.
• In a cash flow approach, the basic scenario doesn’t include the labour cost for the
farmer. While in a farming management scenario, we consider an hourly cost of 7,62
€/h (minimum salary in France). In this last approach, it’s therefore possible to
evaluate the efficiency of the farmer labour.
As it seems impossible to anticipate the future evolution of prices and costs, we assumed
constant values. For instance, a rise or a drop of timber value would respectively increase or
decrease the tree revenue.
Main results
Labour impact for one silvoarable hectare

temps de travail temps de travail
(h/ha/an) (h/ha/an)
16 16
agriculture agriculture
14 cultures intercalaires 14 cultures intercalaires
arbres arbres
12 12
10 10
8 8
6 6
4 4
2 2
0 0
1 6 11 16 21 26 31 36 41 46 1 6 11 16 21 26 31 36 41 46
année année
Case 1: Plantation of 120 trees/ha Case 2: Plantation of 50 trees/ha
Table 64: Labour evolution in the management of a silvoarable plot during the tree
rotation, separating the crop from the tree labour.
An essential condition for adopting agroforestry from the farmers’ point of view is that they
don’t want to devote more time to a new system. If the farmer planted more trees (case 1),
he would need 1 to 1.5 days each year to maintain the trees. But in the second half of the
rotation, the labour decreases progressively due to the fact that trees don’t need more
special maintenance and that the intercrop activity is reduced. If he plants fewer trees, the
impact during the first years is poor. With the small density, the intercrop activity is longer,
because the crop yield is not so affected by the trees. The labour requested in the second
half of the rotation is therefore lower but very near from the initial scenario.
Prediction of yield evolution

Crop yield evolution
Predicting the crop yield during the second half of the rotation is a perilous venture. If we
know the behaviour of the intercrop during the first half thanks to experimental measures on
existing plots, we asked the bio-economics model to predict the yield evolution. In our
simulation, as we said, we used the LER-Safe prediction. We made the essential hypothesis
that the LER must be include between 1 and 1.4. This condition helps us to determine a
possible range of crop yield evolution, from the pessimist one to the optimist one (see Table
129).
Tree plantation Tree Harvest
100 100
Optimist
80 80
Crop yield (%)
60 60
Pessimist
40 40
Intercrop Yield
20 Pure Crop 20
0 0
0 Time
Table 65: Evolution of the relative intercrop yield according to optimist or pessimist
view about the tree competition. Case of one ha of wild cherry with an initial density of
120 trees/ha for a final density of 80 trees/ha.
In this example of a plantation of wild cherry at 80 trees/ha (final density), which means a
distance between the trees rows of 25m, the crop yield represent more than 90% during the
first half of the tree rotation. According to the interaction level, the crop yield varies between
30 and 75 % of the pure crop yield of reference the year before harvesting the trees.
The crop yield depends on different parameters:
• The parameters due to some initial choices: the crop nature (a sunflower will be more
affected by the shadow of the adult trees than a cereal), the density of the plantation
and the distance between the lines, choice of the land unit (a deeper soil will be more
adapted),...
• The parameters depending on the capacity of the farmer: well pruned trees, tree root
maintenance (root cutting), …
In our economical scenarios, we have tested the different level of interaction.
Tree yield evolution

As for the crop yield estimation, we put forward the hypothesis of different level of timber
productivity. But for our simulations, we only use one prediction of timber production. To

validate our approach, we use a very cautious estimation of production (see Table 130). Our
results can therefore be considered as the minimum result we can get from our hypothesis.
140 140
Interval
120 120
Basic
Standing volume (m3/ ha)

100 scenario 100
Pure Optimist
80 plantation 80
60 60
40 40
Pessimist
20 20
0 0
5 10 15 20 25 30 35 40 45
Time
Table 66: Range of timber volume evolution for an initial plantation of 120 wild cherry.
The figure indicates of the cautious hypothesis of standing volume we used for our
simulations (77 m3 for 80 final trees).
Cash flow impact

To evaluate the impact of the project on the cash flow, we must distinguish first the
investment cost and then the evolution of the annual cash flow depending of the crop yield
evolution and the possible over cost to crop between the trees in comparison with a pure
crop system.
Initial investment
The poor number of trees to plant in an agroforestry system reduces considerably the
investment cost if we compare with a current afforestation cost on agricultural land. The tree
cost is nonetheless higher. The owner will choose a better quality of the trees and will have
to protect each one with a strong protection: each tree has a possible future value and
demands a special attention.
The total cost of a plantation (without subsidy) varies between 500 and 1000 euros/ha
according to the tree specie (the walnut plantation being the most expensive). This cost
represents between 20 to 60 % of the average cost in the case of common land afforestation
(see Table 131).

Afforestation
1 233 €/ha
120 trees/ha
Poplar 695 €/ha
367 €/ha
50 trees/ha
1 518 €/ha
Wild Cherry 469 €/ha

267 €/ha
1 633 €/ha
Walnut 1 034 €/ha

517 €/ha
Table 67: Comparison of the investment in agroforestry and forestry scenario,

WITHOUT subsidy.
In France, it’s current to get a subsidy of 40 to 70% to cover the investment cost and the
maintenance cost of the trees during the 4 first years (except in Franche Comté).
Since 2004, the French Government decided to suspend all economic aids to the land
afforestation, except for agroforestry. In our simulations, we decided to conserve this aid, to
be able to compare between the two options (see Table 132).
Afforestation
617 €/ha
120 trees/ha
Poplar 348 €/ha
184 €/ha
50 trees/ha
759 €/ha

134 €/ha
817 €/ha
Walnut 517 €/ha

259 €/ha
Table 68: Comparison of the investment in agroforestry and forestry scenario, WITH
subsidy.
Cash flow evolution
Evolution of the cash flow at the plot scale

The cash flow evolution will depend of the crop yield evolution and the LER level we have
selected and the final density. For example, in the Table 133, we’ve illustrated the cash
evolution for two different densities but for a medium LER level.

100 100
90 50 trees/ha 90
80 to 90 %
80 70 to 85 % 80
70 70
% Annual Gross Margin
120 trees/ha
60 60
50 30 to 60 % 50
40 40
30 30
20 Silvoarable Gross Margin 20

Agricultural Gross Margin
10 10
0 0
Time Trees Harvesting
Plantation
Table 69: Evolution of the annual cash flow for a probable scenario with wild cherry
(LER=1.07 for a density of 50 trees/ha and 1.15 for a density of 120 trees).
Being cautious in our forecast, we notice nonetheless that at half of the rotation, the gross
margin still represent 80 to 90 % of the agricultural gross margin. We must underline that in
our simulations, we’ve considered that the crop payment area is reduced progressively by
the tree area. In case of the silvoarable area was eligible in its totality, the impact on the cash
would be sensible, above all in some regions where man get poor crop yield and where the
crop payment is essential in the gross margin calculation (Franche Comté for example).
Let’s also underline the fact that in the INRA experimental plots, the LER reaches more 1.3
than 1.15 that we have chosen in our simulation with an initial density of 120 trees/ha.
Influence of the CAP payment policy

Inside the first pillar policy, the situation of the agroforestry plots could be different depending
of each country member. In fact, at a European level, the agroforestry plot could be eligible
to the Compensatory Payment. We compare here the possibility to get the payment on the
whole area (Request of the Safe consortium) or only on the intercrop area (French situation).
The impact of the eligibility given to the whole surface on the profitability is not so important.
In all our simulations, the profitability increases by 3% in the best option for agroforestry. The
impact is more at a cash flow level, when the crop gross margin is low. That’s typically the
case for the farms where:
• The crop component is lower than the payment component in the gross margin
calculation (Mediterranean area or farm with high cost of production)

The yield is decreasing faster in the silvoarable scenario (high density of plantation or strong
impact of the trees on the crop RA)
100%
% of the Arable Gross Margin
80%
60%
agriculture
Payment on intercrop area - 50 trees/ha
40% Payment 100% - 50 trees/ha
Payment on intercrop area - 120 trees/ha
Payment 100% - 120 trees/ha
20%
0%
2% 12% 22% 32% 42% 52% 62% 72% 82% 92%
Time (Tree rotation)
→
Table 70: Influence of the different CAP payment policies in agroforestry on the annual
cash flow evolution.
Evolution of the cash flow at the farm scale

At the farm scale, one of the first questions of the farmer is about the importance of the area
to plant. Does he have to plant on a big area? In several plots or in a single plot? There is no
only one answer. According to the strategy of the farmer, a large range of scenarios is
available. The choice will depend to the cash flow context and to know if the farmer can
support a strong investment or not, and above all if he aims to decrease progressively his
crop activity or not. The labour availability is also a strong parameter to decide which area to
plant. According to our simulation and experimental experience, we often recommend not
planting more than 10 % of the cropping area. In that case, the impact on the farm gross
margin is less than 3 % in average on the first half of the tree rotation. A gradual plantation
will allow a reduction of the cash flow impact (see Table 134).
435 %
191% 183%
178% 180%
% of Farm Gross Margin without AF
175 Farm with 8% silvoarable area 175 Farm with 8% silvoarable area
Farm with 100 % of cropping area Farm with 100 % of cropping area
150 150
125 125
100 100
75 75
50 50
0% 20% 40% 60% 80% 100% 120% 0% 20% 40% 60% 80% 100% 120%
% of the tree rotation % of the first tree rotation
a. Case of a single plantation b. Case of a gradual plantation

Table 71: Comparison of the cash flow evolution when the farmer plants 8 % of his
cropping area (50/50 Walnut/Wild cherry). We compare the option where the farmer
would plant the silvoarable area in once time or if he decides to plant 2 % every 5
years during 20 years.
A gradual plantation will also allow a soft distribution of the timber income in the time from
the moment where the owner begins to harvest the first mature trees (case b). From this
moment, the timber income is regular. In our example, he can harvest the trees every 5
years. In this context, the farm gross margin increase by 15 %. According to the importance
of the plantation and of the species he planted, a farmer could increase his farm income

between 10 to 100%. Of course, it can suppose a long term to wait for the farmer before the
first tree harvest…
Profitability of a silvoarable investment
Comparing a silvoarable scenario with agricultural scenario

For our simulations, we have selected 3 kinds of farms:
• Farm with good crop yields and few fixed costs.
• Farm with medium crop yields with few fixed costs.
• Farm with medium crop yields and high fixed costs.
For each farm, corresponding to each region of the LTS of the WP8, we have run different
scenarios according to:
• the tree density: 120 versus 50 for the initial density, which corresponds to a final
density of 80/40.
• the LER level: optimist, probable and pessimist
• the land unit: good/medium
• the 3 tree species: poplar, walnut and wild cherry
108 scenarios have been run in total (36 scenarios / LTS). Table 72 shows a synthesis of the
Agricultural Values for all these scenarios we have calculated for each species according to
the level of LER.
Walnut Wild Cherry Poplar

100%
Agricultural
% of realised simulations
80% Value Index

> 1,35
60% 1,20 - 1,35
1,05 - 1,20
40%
0,95 - 1,04
20% < 0,95
0%
le
le
le
pr st
pr st
pr st
ist
ist
ist
ab
ab
ab
i
i
tim
tim
tim
im
im
im
ob
ob
ob
ss
ss
ss
op
op
op
pe
pe
pe
Scenario for intercrop productivity
Table 72: Profitability of the silvoarable scenarios according to the tree specie and the
LER level.
A first interesting result is that the silvoarable scenarios are at least as profitable as the
agricultural scenario.

Walnut timber is actually the most expensive timber on the market. For a same duration of
rotation, the best results have been logically obtained with the walnut than the wild cherry.
The period of harvesting time is a key parameter in the profitability calculation.
2,00
1,80
1,60
1,40
1,20
1,00
0,80
0,60
0,40
0,20
0,00
Very
Trèswellbien
pruned Bien
Well pruned
formé Badly
Mal pruned
formé
formé (50 ans) (60 ans)
40 years 50 years 60 years
(40 ans)
Table 73: Influence of the maintenance quality on the profitability.

A late in the pruning dates can put the harvesting date back by 10 or 20 years, above all for
some sensitive specie such as the hybrid walnut. In this example, a late of 20 years means a
reduction of 60% of the profitability in comparison of the agricultural profitability.
Influence of the TGA on the Agricultural Value

The value of the Tree Growth Acceleration has a strong impact on the profitability of the
silvoarable scenarios. This impact is stronger for the scenario with higher densities of
plantation. In the following figure, we noticed that the scenario with a density of 120 ha react
much quicker than a scenario with 50 trees.
Again, in our simulations, we used a TGA of 1,20, which could be considered as a pessimist
approach with what we observe in the reality. For example, in the Jollet's case, the
agricultural value would have been increased by 10 to 15 %.

Jollet TGA
1,10
Indice of Agricultural Value

1,05
Hypothesis simulation
1,00
120 wild cherry/ha

0,95
50 wild cherry/ha
0,90
0,8 1 1,2 1,4 1,6 1,8 2
Tree Growth Acceleration
Table 74: Influence of the TGA on the Agricultural Value
Which density to plant to get the best profitability?

A common question from the farmers is about the number of trees to plant. The farmers often
want to maintain a correct crop yield during the whole rotation but trying in the same time to
get the best investment for timber. Other decides to plant more trees with the aim to
decrease the agricultural activity, even till to suppress the intercrop. We didn’t take this case
in this study.
For each species, Walnut, Wild cherry and Poplar, according to our production hypothesis,
we simulated the impact of the density to the LER but also to the Agricultural Value.
1,6
1,4
LER_optimist
LER_medium
1,2
LER_pessimist
Val-agri_optimist
1
Val-agri_medium
Val-agri_pessimist
0,8
0,6
0 20 40 60 80 100 120 140
WILD CHERRY - Finale Density (trees/ha)
Table 75: Influence of the tree density on the LER value and the Agricultural value for
wild cherry, walnut and poplar.
We observe that for each species, the best density to get the optimum LER is higher than the
best density to get the optimum Agricultural Value. For the species with a poor Tree RA
(Walnut and wild cherry), the range of density is similar (see Table 76). The best density
would vary between 80 to 120 trees/ha to get the highest LER, while the farmer will get the

best profitability with a density included between 60 and 90 trees/ha. Of course, with a higher
TGA, this range would increase.
Result Wild Cherry Walnut Poplar

LER 80 - 120 80 - 120 130 - 200
Agricultural Value 60 - 90 60 - 90 100 - 130
Table 76: Range of density to get the optimum LER and Agricultural Value results for
each species (trees/ha – final density).
For the poplar, the optimum densities are higher than for the 2 others species. This result is
due to the fact that the biomass produced by the silvoarable poplar is similar to the biomass
produced by the forestry poplar. The Tree RA is therefore higher for a given density
compared to other species, which demand more important fellings.
What could influence these results? As we already said, the TGA level could strongly
influence these results, giving priority to higher densities. The policy schedule and the price
level of the crop and tree component will be therefore the most important parameters. In the
case of the walnut, the choice of a density of 75 trees/ha is a wise option. Below, the farm
doesn’t want to take any risk at a long-term period, above he bets more on the trees.
Comparing a silvoarable scenario with a forestry scenario

We compare also the case where the farmer was hesitating between a forestry investment
rather than a silvoarable investment from a profitability point of view.
Agricultural Value Index

1,50
of a silvoarable scenario
of a pure plantation scenario
1,00
1,55
1,21
0,50 1,04 1,00
0,89
0,48
0,00
Poplar Walnut Wild Cherry
Table 77: Comparison of the profitability of the silvoarable and afforestation scenario
with the agricultural scenario. Silvoarable plantation of 120 wild cherry by ha
characterized by a LER of 1,15.
In this example, we explore the case of a probable LER of 1.15 in the silvoarable option. In
almost all our simulations, the silvoarable options are more profitable than the forestry option.
The forestry option may be more profitable in the case where the crop margin is very poor,
above all if it’s possible to plant some valuable species such as walnut for example.
It’s also interesting to notice that for the poplar, the silvoarable option could be a possibility to
stimulate the poplar market. In France, the poplar area is currently decreasing because of

the price fall of the timber (less than 45 €/m3). Agroforestry could therefore be a possible
strategy to reduce the market risks.
Property holdings evaluation in agroforestry

According to his age, a landowner who plants trees, will not necessary benefit from the
harvest… But, as a farmer told us, a farmer has three possibilities of income: the sale of his
products, the stock variation and the possibility to make a capital gain. In this last option, a
silvoarable plot is a capital, which could be evaluated if necessary (inheritance, expropriation,
etc). The land evaluation in agroforestry is the combination of the agricultural land evaluation
and the future value of the trees.
16 000 €
No commercial value with commercial value
14 000 €
12 000 €
10 000 €
Euros by ha
8 000 €
6 000 €
4 000 €
2 000 €
0€
10 20 30 40
Age of the trees (years)
Agriculture agroforestry
Table 78: Evolution of the monetary value of the silvoarable land according to the age
of the trees. In agroforestry, this value is the sum of the agricultural value plus the
timber future value. If the young trees could have a future value, for example at 10
years old, they don’t necessary have a commercial value in the sense that the
landowner can not expect some income if he cut them.
In this example of a wild cherry plantation, the capital evaluation may represent between
twice and four time the agricultural land value according to the age of the trees. In the case
of a walnut plantation, it may represent till 7 times this value 10 years before the tree
harvesting.

Photo 1: In this plot of 4 ha, the wild cherries are 30 years old. The value of the
standing volume is estimated to 4 000 €/ha, which represents the same value of the
agricultural land. But the future value of this plantation is much higher and overpass
the 10 000 €/ha.
Main conclusions
To invest in agroforestry represents a light investment in money and labour comparing with
some new systems of diversification. In our simulations, the profitability reaches 10 to 50 %
with walnut, and -5 to +15 % with wild cherry and poplar, comparing with the agricultural
scenario.
A regular calendar of plantation on a few surfaces is a good option for the farmer (labour and
cash flow impact). 10 % represents between 2 and 3 % of reduction of the farm gross
margin. But in the balanced period, the income increases by more than 15% (mixed
plantation of walnut and wild cherry trees). The gross margin could double if the farmer
plants progressively his whole cropping area. But in that case, it means a stronger impact on
the initial cash flow and demands a consequent labour...
If the best bio-physical option is to plant between 80 to 120 trees by hectare (130 to 200 for
the poplars), the best economical option is to plant a lower density around 60 to 90 trees by
hectare (100 to 130 for the poplar). This means a distance between the trees lines varying
between 24 to 36 m.
All our simulations haven’t taken into account the environmental benefits such the carbon
sequestration, or the impact on the nitrogen pollution. These aspects could be calculated and
to be summed to the whole profitability of the silvoarable systems.
Bibliography
Borrell, T. (2004) De l’importance des interactions arbres-cultures sur les performances

économiques de l’agroforesterie tempérée. Mémoire de Diplôme d’Agronomie Approfondie,
ENSAM-INRA, Montpellier. 98 p + annexes
Boulet-Gercourt, B. (1997) Le merisier. IDF, 2ème édition. 128 pp.
Coulon F, Dupraz C., Liagre F., Pointereau P. (2000) Etude des pratiques agroforestières
associant des arbres fruitiers de haute tige à des cultures et pâtures, Rapport au ministère
de l’environnement, 199 p, Solagro/INRA, Fr

CRPF (1997) Boiser une Terre Agricole. 28 pp.
Dupraz C., Lagacherie M., Liagre F., Boutland A., (1995). Perspectives de diversification des
exploitations agricoles de la région Midi-Pyrénées par l’agroforesterie. Rapport de fin d’étude
commandité par le Conseil Régional Midi-Pyrénées, Inra-lepse éditeur, Montpellier, 253 pp.
Dupraz C., Lagacherie M., Liagre F., Cabannes B., (1996). Des systèmes agroforestiers pour
le Languedoc-Roussillon. Impact sur les exploitations agricoles et aspects
environnementaux. Inra-Lepse éditeur, Montpellier, 418 pp.
Dupraz, C., Liagre, F. & Borrell, T. (2005) The Land Equivalent Ratio of a silvoarable
agroforestry system. In preparation.
Graves, A.R., Burgess, P.J., Liagre, F., Dupraz, C. & Terreaux, J.-P. (in preparation) The
development of an economic model of arable, agroforestry and forestry systems. To be
published soon in Agroforestry Systems.
IDF (1997) Les noyers à bois. 3ème édition, Février 1997. 132 pp.
Liagre F., (1993). Les pratiques de cultures intercalaires dans la noyeraie fruitière du
Dauphiné. Mémoire de Mastère en Sciences Forestières, ENGREF, Montpellier, 80 pp
Segouin O., Valadon A., (1997) Enquête sur les boisements récents de peupliers en Lot-et-
Garonne, Analyse de pratiques agroforestières ; les cultures intercalaires. Cemagref,
Nogent-sur Vernisson, 45 pp.
Souleres, G. (1992) les milieux de la populiculture, IDF, 310 pp.
Terreaux, J.-P. & Chavet, M. (2002) Problèmes économiques liés à l’agroforesterie :

éléments qualitatifs et quantitatifs. Silvoarable Agroforestry For Europe (SAFE) ; Cabinet
Michel Chavet, Paris – UMR Lameta, Montpellier.
Vandermeer, J. (1989) The Ecology of Intercropping, Cambridge University Press, 225 pp.
Annexes
Annex 1: Detailed description of the LERbased-Generator
Principle
Farm-sAFe does not have any biophysical module, the time-series must be generated
independently: pure crop and intercrop yields, timber production in forestry and agroforestry.
We used a generator constrained by the LER-biomass: depending on a previously fixed
value and on a quite low number of parameters, these times-series are produced. The
starting and final points are known, the evolution between them is drawn thanks to a logistic
equation.
A key characteristic is that the climatic variability is not taken into account. It would have
necessitated defining the impact of variables (temperature, water, light, etc.), which are not
implicated in this type of constrained prediction. Nevertheless, we assume that except in very
particular cases, this variability does not have any impact on the economic results: on a
whole revolution (20 to 60 years), “bad weather” years are compensated by “good weather”
ones, as we are not interested in year-by-year results but final profitability and global
evolution of financial results. Because of discounting, climate would only have a strong effect

if “bad weather” years were concentrated in a specific part of the revolution, which is very
unlikely to happen.
Notations
We use the words “forestry” and “forest trees” for all types of pure trees plantation, even
when the initial density is very low, such as for walnut.
A distinction is made, in forestry and agroforestry, between the trees which are cut at
thinnings and the trees which are maintained until last felling: the firsts are called “thinned
trees”, the others “felled trees”.
We call “timber” the bole of the tree, which has the highest commercial value. The same
word is used for the thinned trees, even if the bole is often too small to be sold as good
timber.
VF is the individual forest tree timber volume at forestry reference density.
VAF is the individual agroforest tree timber volume, depending on the density.
VmaxAF is the maximum individual agroforest tree timber volume.
Parameters
• Parameters per tree species
- DC, the critical agroforestry density, i.e. the density at which the tree growth
potential is attained: the individual agroforest tree timber volume is equal to VmaxAF,
the trees cannot be bigger, even at lower densities.
- TGA, the coefficient of individual Tree timber volume Growth Acceleration in

low density AF conditions, or Tree Growth Acceleration; e.g. 1,2 indicates that the
individual agroforest tree timber volume at a lower or equal density than DC will be
20 % bigger than the one of a forest tree, due to the positive impact of both low
density and intercropping (exceeds of nitrogen, less competition than the perennial
vegetation classically established between forest trees, etc…).
- Timber To Biomass in forestry, e.g. the timber contribution to the total biomass
of a young forest tree (TTByoung-F) and of a felled forest tree (TTBfell-F);
- Timber To Biomass of a felled agroforest tree (TTBfell-AF);
- maximum value for the forestry ratio: biomass of all the thinned trees/biomass
of all the felled trees;
- individual tree timber volume of the agroforest tree at thinning.
• Parameters of the logistic curves
- curvature and inflexion for the individual forest tree timber volume, for the
individual agroforest tree timber volume;
- curvature and inflexion for the height of the forest trees, of the agroforest
trees;
- curvature for the intercrop yields.

• Parameters used only as Farm-sAFe entries2
- final tree height (same in forestry and agroforestry);
- maximum bole height (same in forestry and agroforestry);
- fixed value of the ratio: firewood volume/timber volume in forestry, in

agroforestry.
Entries
- arable rotation, reference yield and threshold yield for profitability for every
crop;
- tree species, revolution duration (60 years maximum);
- forestry: reference production, initial density, years of thinnings (maximum 5)

and numbers of thinned trees;
- agroforestry: initial density, number of trees cut in the unique thinning, plot
design (distance between tree lines, initial width of the intercropped alley, width of
the intercropped alley reduction step);
- LER-biomass aimed.
The generation of data-sets
• The first step is the generation of the time-series of the monocropping systems:
The time-series of pure agricultural yields are produced simply by repeating the reference
yields as many times as necessary to last the duration of the revolution.
The time series of the timber production of the felled trees are generated with the following
logistic equation:
Y initial −Y final
Yt = +Y final
  t 
courbure

1 +  T inf lexion  
   
where Yt is the value of Y at year ;
Y initial and Y final are the initial and final values of Y;
T inflexion is the inflexion date (end of linear growth).
Y initial is equal to zero and Y final to any value, as the curve is then distended in order to go
through a point {X’;Y’}: X’ is the date of fell of the trees and Y’ is the reference production (in
m3/ha of timber at felling).
For forestry, 3 other time series are generated :
2
These three parameters are not used in the generation of tree production data-sets (timber), but they
are needed as entries for Farm-sAFe (tree height and production of firewood).

- The timber production of the thinned trees3, as it is considered that they can
be smaller than the felled trees. The same logistic equation is used, with an Y’
calculated in function of 2 constraints:
(i) thinned tree timber volume ≤ felled tree timber volume.
(ii) the parameter “maximum value for the ratio: biomass of all the
thinned trees/biomass of all the felled trees”
- The biomass of both the felled and the thinned trees, thanks to the ratio
Timber To Biomass. We assume that if TTBF may vary with time t, it is a linear
variation:
TTB fell − F − TTByoung − F

TTB F (t ) = × t + TTByoung − F
T
where T is the revolution duration.
The biomass of the felled trees and of the thinned trees is thus calculated by dividing their
respective timber time series by TTBF(t).
• The Relative Areas calculated in function of the Tree Growth Acceleration:
The coefficient of individual Tree timber volume Growth Acceleration in low density AF
conditions, or Tree Growth Acceleration (TGA), permits the calculation of the agroforest tree
timber volume:
- At D ≤ DC, VAF = VmaxAF
- At D = DF, VAF = VF
- D > DF is not possible
VF − V max AF V max AF −VF

- At DC < D < DF, VAF = VF + D × − DF ×
DF − DC DF − DC
Where: D, DC and DF are the actual density, the critical density and the forest
density
VAF, VF and VmaxAF are the actual agroforest tree timber volume, the forest
tree timber volume and the maximum agroforest tree timber volume, with
VmaxAF = TGA x VF
On the basis of the final densities in forestry and agroforestry, the forestry RA- products can
then be deduced.
With TTBfell-AF, we easily know the agroforest trees biomass at felling.
With regards to the low initial density in agroforestry, we assume that the thinned trees grow
as well as the felled trees: the thinning is early enough to avoid a strong effect of competition,
3
There is only one time series for all the thinnings: late-thinned trees have the same rate of growth as
early-thinned trees.

thus the individual thinned tree timber volume is the same as the one of a felled tree at that
time. And as the number of thinned trees is low and the thinning quite early, the volume of
thinned biomass is poor enough to permit us to consider a fixed TTBAF in time. Thus the
volume of thinned biomass in agroforestry is calculated by dividing the thinned timber
production by TTBfell-AF.
The forestry RA-biomass can then be calculated.
The arable RA-biomass is deduced in function of the aimed LER-biomass. It is equal to the
arable RA-products, as we assume that the proportion of grain in the biomass of the crop is
the same in agriculture and in agroforestry.
• The generation of the agroforestry data-sets:
The agroforestry timber time-series are generated with the same logistic equation: Y’ is then
the forestry reference production multiplied by the forestry RA-products.
Until the thinning, the volume of timber of the thinned trees is taken into account simply by
adding the equivalent number of trees with the same individual tree timber volume.
The intercrop time-series are also generated with this logistic equation, with an Yfinal equal
to 0: one time-series per crop of the rotation (maize, wheat, oilseed, etc...). For each crop,
the curve is adjusted in function of the threshold yield and the width of the intercropped alley
reduction step: as the yield per total ha decreases with time due to tree growth and
increasing light competition, we assume that the cropped area is reduced by successive
steps (see Table 79). A reduction of the width of the intercropped alley happens every time
the yield per cropped ha passes under an economically defined threshold. The last reduction
corresponds to the suppression of the intercrop.

w
Table 79: chronological schema of the intercropped area in the alley in function of tree
growth.
The width of the intercropped strip (w) is reduced at t1 and t2 in order to increase the
mean yield per cropped ha. Its next reduction at t3 corresponds to the suppression of the
intercrop. In the generator, up to 6 reductions can be made.
The intercrop curves are adjusted by modifying their inflexion date so that the sum of the
intercrop productions is equal to the crop reference yield multiplied by the arable RA-
products.
The intercrop time-series are then mixed according to the arable rotation to obtain a single
time-series.
• The same tree height curve time series in forestry and agroforestry
A last time-series is generated for both forest and agroforest trees : their height growth. It is
not used in the timber volume calculation, but this time-series is needed in Farm-sAFe for its
“autoprune” function.
We use the Boltzmann logistic equation :
Yinitial − Yfinal
Y(t) =  t −Tinf lexion 
+ Yfinal − Y(t = 0)
1+ e  curvature 
As for the timber time-series, Y initial is equal to zero and Y final to any value, as the curve is
distended in order to go through a point {X’;Y’} : X’ is the date of fell of the trees and Y’ is the
aimed height.

Annex 2: Labour, revenues and costs in the 3 types of farms
Assignabl
Single Farm Variable
Price e fixed
Annual labour Payment costs
Crop costs
(h/ha)
(€/t)
(€/ha) (€/ha)
(€/ha)
110
wheat 6,
(straw 30 €/t) 343 300 207
Hy-Lc
oilseed 5,5 215 360 302 207
set aside 1,5 – 338 15 207
110
wheat 6
(straw 30 €/t) 345 302 252
oilseed 5,5 220 361 391 252

Ly-Lc
sunflower 5,5 280 361 262 252
set aside 1,5 – 345 27 252
102,10
wheat 7
(straw 30 €/t) 328 278 315
oilseed 5,5 220 348 390 315

Hy-Hc
maize 7 85,4 348 434 315
set aside 1,5 – 328 15 315

Annex 3: Economic data relative to monocropped or intercropped walnut, wild
cherry and poplar in the 3 farms
Tree timber standing value
Standing value (€/m3)
Walnut Wild cherry Poplar

Tree timber volume
(m3/tree)
thinned trees felled trees thinned trees felled trees felled trees
0.03 0 0 0 0 0
0.04 10 10 10 10 7
0.05 10 10 10 10 7
0.06 10 10 10 10 7
0.07 10 20 10 10 7
0.08 10 20 10 10 7
0.09 10 30 10 10 7
0.1 10 40 10 10 7
0.11 10 60 10 10 7
0.12 10 80 10 10 8
0.13 10 100 10 10 8
0.14 10 126 10 10 8
0.15 20 135 10 10 8
0.16 20 144 10 10 8
0.17 20 153 10 12 8
0.18 20 162 10 15 8
0.19 20 171 10 20 13
0.2 30 180 10 40 15
0.3 30 270 15 55 20
0.4 40 360 20 75 24
0.5 50 450 22 150 28
0.6 100 540 25 250 32
0.7 190 630 35 275 35
0.8 220 720 45 300 37
0.9 300 810 55 325 39
1 400 900 65 350 41
1.1 500 925 75 360 43
1.2 600 950 85 370 45
1.3 700 1000 95 380 46
1.4 800 1000 105 380 47
1.5 900 1000 115 380 48
1.6 1000 1000 125 380 49
1.7 1000 1000 135 380 50
1.8 1000 1000 145 380 51
1.9 1000 1000 165 380 52
2 1100 1100 175 380 53
3 1200 1200 200 380 55
4 1300 1300 200 380 55

INRA Report
Establishment costs
Cost of Labour
Labour for Labour Labour for Labour for Labour for
Cost of individual for
ground for full planting tree localised
plant tree marking
preparation weeding trees protection weeding
protection out
(€/tree) (€/tree) (h/ha) (h/ha) (h/ha) (min/tree) (min/tree) (min/tree)

agroforest walnut 6.00 1.50 4.00 0.50 7 2 2 0.50
forest walnut 4.00 0.50 6.50 1.50 4 2 1 0.50
agroforest wild cherry 1.00 1.50 4.00 0.50 7 2 2 0.53
forest wild cherry 0.50 0.50 6.50 1.50 4 2 1 0.53
agroforest poplar 4.00 0.50 12.00 0.50 7 2 2 0.50
forest poplar 4.00 0.50 16.00 1.50 4 2 1 0.50
Maintenance and pruning costs
Labour for Materials for

Annual Annual cost Height at Minutes per Minutes per
Weeding annual grass annual grass Height at Removal of
labour for of herbicide first tree at first tree at last
period sward sward last prune pruning
weeding for weeding prune prune prune
maintenance maintenance
(years) (min/tree) (€/tree) (h/ha) (€/ha) (m) (min/tree) (m) (min/tree) (min/tree)
agroforest walnut 1-3 0.50 0.14 2.0 30 1.00 1.00 4 7.00 4.00
forest walnut 1-3 0.50 0.14 4.0 90 1.00 0.20 4 7.00 4.00
agroforest wild cherry 1-3 0.50 0.14 2.0 30 1.00 1.00 6 6.40 4.00
forest wild cherry 1-3 0.50 0.14 4.0 90 1.00 0.18 6 6.40 4.00
agroforest poplar 1-3 0.53 0.14 2.0 30 1.50 1.00 8 10.00 4.00
forest poplar 1-3 0.53 0.14 4.0 90 1.50 1.00 8 10.00 4.00
Labour for thinning and felling
Thinnings Clear felling

Marking up &
Removal of tree Labour Removal of tree
labour
(min/tree) (min/tree) (min/tree) (min/tree)
agroforest walnut 7 5 4 2
forest walnut 7 5 4 2
agroforest wild cherry 7 5 4 2
forest wild cherry 7 5 4 2
agroforest poplar 7 5 4 2
forest poplar 7 5 4 2
Administrative costs
In agroforestry, the land tax is the same as in an agricultural plot. It was thus applied to
the tree strips.

INRA Report
Land tax Insurance

(€/ha) (€/ha)
Hy-Lc AF plot (agricultural tax) 44 20
(Centre) forestry plot 30 20
Hy-Hc (Franche- AF plot (agricultural tax) 52 20
Comté) forestry plot 36 20
Ly-Lc (Poitou- AF plot (agricultural tax) 58 20
Charentes) forestry plot 39 20
WP10. Project co-ordination

INRA-System administered the SAFE web site during the entire SAFE project. The
SAFE web site popularity increased a lot since the last 6 months, thank to a good
communication on the project (extension papers, member participation on international
conferences etc... )
Table 80: Safe web site public section visits progression

from May 2002 to January 2005
These documents have been published on the public section of the safe web site under
“Agroforestry > extension papers” and “Safe project > Scientific publications”

INRA Report
Table 81: French extension paper list on the Safe web site
French English Spanish German

Extension papers 22 3 3 4
Scientific Publications 11 1
Oral presentations 6
Posters 7
Table 82: Number of documents, by languages, under the public section of safe
web site

INRA Report
Visitor Number of hits

GOOGLE 4259
INRA - Montpellier 549
INRA - Others 502
Searching engine 477
CIRAD 367
French Ministry of Agriculture 155
University of Florida 142
ENITAB 135
Cemagref 123
Napier University - Edinburgh 102
CNRS - INST 81
Dumrath & Fassnacht KG 76
Chambre Agriculture Ille et Vilaine 72
Ecole Supérieure Agriculture Angers 71
Universidad Complutense de Madrid 65
University of Central Lancashire 64
CNRS 62
Université du Havre 63
Universitaet Hohenheim 53
Commission of the European Communities 50
Faculte des Sciences agronomiques de Gembloux 48
Forest Research Institute Budapest 47
SPIEGEL 45
Universidad Politecnica de Madrid 43
Rectorat de Bordeaux 42
ENFA Toulouse 41
Universite de Laval 41
Université de Moncton 38
CTIFL 37
Agropolis - Montpellier 35
Bulgarian Academy of Sciences Network 35
Chambre départementale agriculture du Doubs 35
University Bergakademie Freiberg 33
Swiss Federal Institute of Technology Zurich 32
Rectorat de académie de Bordeaux 32
ACTA Informatique 28
ENSAT 45
TU Bergakademie Freiberg 28
Conseil général de Hérault 26
Universität Hannover 26
Universidad de Burgos 26
Universite Catholique de Louvain 25
Conseil Regional du Languedoc Roussillon 25
University of Maribor 24
Universite Montpellier I 22
Table 83: 50 most frequent visitors (except consortium members) of the safe
web site since March 2003

INRA Report
2 Contractor 1: INRA-AMAP
Partner: 1: INRA-AMAP
Institute: AMAP: UMR Botanique et Bioinformatique de l’Architecture des Plantes;

TA 40/PS2; 34398 Montpellier cedex 5
Scientific team
Name Tel Fax e-mail
Auclair Daniel 33 467 61 65 34 33 467 61 56 68 auclair@cirad.fr
Caraglio Yves 33 467 61 65 48 33 467 61 56 68 caraglio@cirad.fr
de Coligny François 33 467 61 71 68 33 467 61 56 68 coligny@cirad.fr
Dauzat Jean 33 467 61 65 76 33 467 61 56 68 dauzat@cirad.fr
Sabatier Sylvie 33 467 61 65 85 33 467 61 56 68 sylvie-
annabel.sabatier@cirad.fr
Parveaud Claude-Éric 33 467 61 65 85 33 467 61 56 68 parveaud@cirad.fr
Within the UMR AMAP (joint research unit), Daniel Auclair (DR2), François de Coligny
(IR2) and Claude-Éric Parveaud (BTH) are INRA scientists, whereas Yves Caraglio,
Jean Dauzat and Sylvie Sabatier are members of CIRAD. Several other members of
AMAP have partly contributed both to the botanical aspects of the project (Éric Nicolini)
and the software development (Frédéric Boudon, Jean-François Barczi, Christophe
Godin, Christophe Pradal). Several tools used within the SAFE project are still under
development.
Objectives
The main role of AMAP was to build detailed simulation mock-ups of trees, to help
calibrate the tree-crop models. Detailed mock-ups of three important tree species for
SAFE were to be provided, based on measurements in experimental situations. These
mock-ups were then to be used as reference descriptors of use of space by trees, in
order to test the validity of the Hi-sAFe model. The species to be studied in detail are
Prunus avium, Populus sp., and Juglans nigra x regia.
AMAP methodology has been briefly described in the first year report. In the first year,
data collection and parameter estimations had been performed for Wild Cherry and
Hybrid Walnut. The main results of the second year concerned AMAPsim parameters. In
the third year, the parameters for AMAPsim tree simulations were improved, and the 3–
D tree mock-ups were used for validation of more general models.
In this last period the results obtained during the first three years were used i) for
estimating the effect of climatic variations on growth fluctuations of Hybrid Walnut, and ii)
to estimate parameters for Hi-sAFe and validate the mock-ups. The final reports for
Milestone 7: “Mock-ups of virtual trees for 3 key species : Prunus avium , Populus ,
Juglans nigraxregia” were published.

INRA Report
Time spent on the different workpackages

Time spent on the different workpackages during the fourth year
Name WP4 WP6 Total

Auclair Daniel 1.9 1.9
Caraglio Yves 0.2 0.2 0.4
de Coligny François 0.0
Dauzat Jean 0.2 0.3 0.5
Sabatier Sylvie 0.2 0.2
Parveaud Claude-Éric 3.2 3.2
Total 0.6 5.6 6.2
Contribution to workpackages
WP4. Above-ground interactions (0.6 person-months)
Climate effect on the growth of hybrid Walnut (Juglans nigra x regia )
The aim of this study is to estimate the climatic intra- and inter-annual variations on the
inter-annual fluctuations of the growth of Walnuts and the synchronism of individuals, to
improve the hybrid Walnut mock-ups.
A sample of twenty-six 9-year-old hybrid Walnut trees that grew in the agroforestry
experimental area of Restinclières (Prades-le-Lez, Hérault) were analysed. For each
successive annual shoot along the main stem (produced from 1994 up to 2003), the
length, leaf number, diameter, branch number and basal diameter, and tree ring area
have been collected. These data are recorded as multivariate sequences. The climatic
data, minimum and maximum temperature, rainfall and global radiance, are studied at a
daily scale. The statistical method of analysis is based in the combination of a
multiphasic modelling and a linear mixed modelling, following the method described by
Véra et al. (In: 4th International Workshop on Functional-Structural plant Models, 7-11
June 2004, Montpellier, France). This method allows to relate primary growth (shoot
extension) and secondary growth (tree ring) to climatic and ontogenetic factors. The final
analysis and results have been delayed, due to the maternity leave of the principal
investigator (Sylvie Sabatier): they will become available shortly.
WP6. Modelling (5.6 person-months)

The AMAP team contributed to the above-ground modelling by using detailed 3-D tree
mock-ups to estimate allometry parameters within the tree crown, to feed into the
general models:
• Verification of the “pipe-model” theory (Shinozaki et al. 1964);
• Estimation of the volume occupied by the leaves according to the crown

representation scale;
• Estimation of the proportion of woody volume of Hybrid Walnuts between stem

and branches.

INRA Report
The results are presented in Appendix (in French). The general conclusions are the
following:
• The “pipe-model” theory is verified at the global scale, with a regression

coefficient similar for the three trees studied. But there are differences at the
branch or branchlet scale, within individual tree crowns, which could be related to
position in the crown, and/or to branch age. No simple relation could be found at
this scale.
• According to the simplification scale, the estimated volume occupied by the

leaves can vary by a factor 2: a volume at the whole crown scale can be twice
the estimated volume at the scale of an annual growth unit, the branch scale
being intermediate. This ratio can increase with time, in conjunction with the
larger space unoccupied by leaves. Differences between trees can also be
explained by the proportion of short shoots relative to long shoots, which is highly
variable in Hybrid Walnut.
• For a total woody volume of respectively 0.03 m3 / 0.04 m3 / 0.10 m3 , the

proportion of volume in the bole (free of branches) was 0.27 / 0.31 / 0.24.
Dissemination
Participation in meetings and workshops
The AMAP team pursues its implication in various workshops and scientific research
groups: INRA “Club modélisation” ; CAPSIS community ; INRA-EFPA Growth and
Dynamics group ; INRA ADD (Agriculture and sustainable development) ; SEAMLESS:
System for Environmental and Agricultural Modelling - Linking European Science and
Society.; SAFODS: Smallholder Agroforestry Options for Degraded Soils ; IUFRO group
1.15.01 (temperate agroforestry).
Scientific publications
Parveaud C.E., Chopard J., Auclair D. 2005. Reconstruction of foliage geometry on 3D
virtual plants in order to compute radiative balance. In: XXII IUFRO World Congress, 08-
13/08/2005, Brisbane, AUS. Poster session 056 "Modelling forest production - does
scale matter?". Poster paper, accepted.
Several scientific publications are in preparation.
Utilisation de maquettes architecturales de Noyers (in French)

hybrides pour Hi-sAFe. (in French)
By Claude-Eric Parveaud

INRA Report
Table 84: Représentation de la couronne du noyer 162 à 9 ans (année 2003) à

l’échelle foliaire. Les feuilles bleues sont portées par des pousses d’une longueur
inférieure ou égale à 2cm. Un carreau au sol représente une surface de 1m²
A . Vérification de l’hypothèse du pipe model sur trois noyers hybrides âgés de 9

ans
1. Introduction
Le pipe model est l’une des méthodes les plus communément utilisée pour distribuer les
ressources entre le feuillage et la structure ligneuse dans les modèles mécanistes
(« process-based model »). A l’origine, il s’agit d’un modèle purement morphologique.
L’hypothèse du pipe model (Shinozaki et al. 1964) considère qu’il existe une relation
linéaire (de pente η) entre la biomasse foliaire ou la surface foliaire et la surface
conductrice qui alimente l’ensemble des feuilles considérées.

INRA Report
Un grand nombre d’articles montrent que cette hypothèse est parfois vérifiée, parfois
réfutée. La conclusion de ces travaux empiriques est que la réponse dépend de l’espèce
considérée (Sievänen et al. 2000). Différentes améliorations ont été apportées à cette
hypothèse. Dans certains modèles, le paramètre η est une variable qui dépend de la
hauteur dans l’arbre. Berninger et al. (1995) propose d’écrire η comme une fonction de
l’évapotranspiration potentielle et teste cette relation sur des pins sylvestre situés le long
d’un gradient nord-sud en Europe (Berninger et al. 1995), (Berninger et al. 1997).
L’objectif de ce travail est de vérifier si (i) les noyers hybrides répondent à l’hypothèse
du pipe model et (ii) de préciser cette relation en fonction de l’ordre de ramification
considéré (branches, rameaux).
Pour répondre à cette question, nous utiliserons trois noyers dont nous avons mesurés
la topologie et le diamètre basal des pousses annuelles en 2003. L’utilisation de ces
trois maquettes architecturales est possible car (i) nous avons les données nécessaires
pour calculer η et (ii) les surfaces foliaires des maquettes ne sont pas décrites comme
fonction du diamètre des pousses porteuses. En effet, si tel était le cas, tester cette
hypothèse reviendrait à tester ce que nous introduisons comme paramètre à l’entrée du
modèle pour générer les maquettes, ce qui n’a aucun intérêt.
2. Matériels et Méthodes
2.a. Matériel biologique
Trois noyers (tableau 1, figure 1) ont été mesurés en 2003 et 2004 afin de reproduire les
arbres tels qu’ils étaient durant la saison de croissance 2003. Ils sont âgés de 9 ans en
2003.
Nota : Par la suite, ne pas tenir compte du suffixe *P03 dans les figures.
Tableau 1 : Caractéristiques divers des trois noyers hybrides étudiés en 2003.
3311 3610 162

Parcelle forestière forestière agroforestière
Hauteur (m) 5.08 4.95 7.28
Nombre de feuilles 2371 2502 4973
Surface foliaire totale (m²) 61.0 66.8 128.3
Année de première floraison 2003 2001 2001
nombre pousses courtes /
0.07 0.39 0.07
nombre de pousses longues

INRA Report
Table 85 : Représentation 3D des noyers hybrides 3311, 3610 et 162 en 2003. Le

trait noir représente une hauteur de un mètre. Les feuilles positionnées sur des
pousses longues sont vertes, les feuilles positionnées sur des pousses courtes
(longueur ≤ 2cm) sont bleues.
2.b. Modélisation de la surface foliaire des maquettes architecturales
La surface (Sf) des symboles feuilles est fonction de la longueur de la pousse (L) et du
rang relatif de la feuille (r) sur la pousse. Sf est exprimée en cm² et est telle que :
si L [cm] ≤ 2.0 cm alors Sf(r) = 140.0
sinon Sf(r) = 442.7r3 – 1255.9r² + 876.9r + 164.2
Donc ∀ r ∈ [0 ; 1], Sf(r) ∈ [140.0, 344.9]
Cette relation a été obtenue à partir de mesures de surface foliaire (feuilles numérisées
puis analysées avec Optimas 6.5) réalisées sur un échantillon de 108 feuilles en 2003.
2.c. Représentation des maquettes et extraction des résultats
Les maquettes sont construites grâce au logiciel AMAPmod (Godin et al. 1997) à partir
d’informations topologiques et géométriques issues de la digitalisation des structures
ligneuses (Godin et al. 1999).
3. Résultats
3.a. Vérification de l’hypothèse du pipe model à l ‘échelle des branches
La figure 2 représente la somme de la surface foliaire portée par chacune des branches
en fonction de la surface de la section de ces branches. Les trois individus sont
représentés. Les droites sont des régressions linéaires effectuées sur les nuages de
points en conservant l’intégralité des données.

INRA Report
14
3311P03 y = 0.44x
3610P03 2
12 162P03
R = 0.91
Total leaf area (m²)
Linéaire (162P03)
10 Linéaire (3610P03) y = 0.41x
Linéaire (3311P03) 2
R = 0.93
8
y = 0.38x
6 2
R = 0.96
4
0
0 5 10 15 20 25 30
Cross sectional area (cm²)
Table 86 : Somme de la surface foliaire portée par les branches en fonction de la

surface de la section des branches. Une droite de régression est tracée pour
chaque individu.
n (3311P03) = 50 ; n (3610P03) = 38 ; n (162P03) = 30
A l’échelle des branches, la régression linéaire entre la surface foliaire totale et la
surface de la section des branches pour chaque individu est de bonne qualité (r² ≥ 0.91).
La valeur de η (la pente) est proche entre les trois individus.
3.b. Vérification de l’hypothèse du pipe model à l’échelle des branches – individu 3311
La figure 3 est identique à la figure 2 mais ne représente que l’individu 3311. L’âge de la
pousse annuelle sur laquelle s’insère la branche est différencié par des couleurs
différentes.

INRA Report
7 y = 0.41x ; R 2 = 0.92
y = 0.35x ; R 2 = 0.98
6
y = 0.34x ; R 2 = 0.97
5 y = 0.30x ; R 2 = 0.97
y = 0.39x ; R 2 = 0.77 1998
4 1999
2000
2001
3
2002
Linéaire (1998)
2 Linéaire (1999)
Linéaire (2000)
1 Linéaire (2001)
Linéaire (2002)
0
0 2 4 6 8 10 12 14 16

Table 87 Somme de la surface foliaire portée par les branches en fonction de la
surface de la section des branches. Individu 3311. Une droite de régression est
tracée pour chaque âge des branches. L’âge est celui de la pousse annuelle sur
laquelle s’insère la branche.
n (1998) = 9 ; n (1999) = 9 ; n (2000) = 11 ; n (2001) = 5 ; n (2002) = 16.
3.c. Vérification de l’hypothèse du pipe model à l’échelle des branches – individu 3610
La figure 4 est identique à la figure 3 mais ne représente que l’individu 3610. Les
branches situées sur la pousse 2002 du tronc ne sont représentées que par 2 individus
(floraison terminale).

INRA Report
9
Total leaf area (m²) 2
y = 0.44x ; R = 0.98
8 y = 0.47x ; R2 = 0.99
y = 0.32x ; R2 = 0.87
7 2
y = 0.43x ; R = 0.85
2 1998
6 y = 0.37x ; R = 0.99
1999
5 2000
2001
4 2002
Linéaire (1998)
3
Linéaire (1999)
2 Linéaire (2000)
Linéaire (2001)
1
Linéaire (2002)
0
0 2 4 6 8 10 12 14 16 18 20
n (1998) = 11 ; n (1999) = 4 ; n (2000) = 12 ; n (2001) = 8 ; n (2002) = 2.
3.d. Vérification de l’hypothèse du pipe model à l’échelle des branches – individu 162
La figure 5 est identique à la figure 3 mais ne représente que l’individu 162. Les
branches situées sur la pousse 1997 du tronc ne sont représentées que par 2 individus.

INRA Report
14
2
y = 0 .24x ; R = 0.65
12 2
y = 0.48x ; R = 0.90
2
y = 0.40x ; R = 0.94
10 2
y = 0 .46x ; R = 0.95
1997
6 1998
1999
2000
4 2002
Linéaire (1997)
Linéaire (1998)
2 Linéaire (1999)
Linéaire (2000)
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26

n (1997) = 2 ; n (1998) = 7 ; n (1999) = 9 ; n (2000) = 10.
4. Discussion
Pipe model à l’échelle des branches
La figure 2 suggère que l’hypothèse du pipe model se vérifie à l’échelle des branches
sur les trois noyers étudiés. La pente η de la régression linéaire est très proche entre les
trois noyers. Bien que ces arbres aient des profils de couronne différents (figure 1) et un
ratio « nombre de pousse courtes/nombre de pousses longues » différent, il est
intéressant de constater que la pente η de la régression linéaire est très proche entre
ces trois individus.
Pipe model en fonction de la position des branches dans la couronne
Lorsque l’on calcul η en fonction de la position des branches sur le tronc, on observe
des réponses différentes en fonction des individus. Sur l’individu 3311 (figure 3), η
diminue avec l’âge de la branche : pour une même surface de section de branche, les
branches hautes portent moins de surface foliaire que les branches basses.
La relation entre η et la position des branches sur le tronc est plus confuse sur les
individus 3610 et 162. Sur le 3610, les valeurs de η sont similaires quelque soit la
position des branches sur le tronc, sauf pour les branches situées sur 2000 où η est plus
faible (η = 0.32). Deux valeurs (entourées en bleues) semblent à l’origine de cette
différence. Quelle explication ? Rester prudent dans l’interprétation de cette différence
car le nuage est plus étalé que pour les autres années (r² plus faible). Quant au 162, les

INRA Report
valeurs de η oscillent entre 0.40 et 0.48 mais sans relation directe avec l’âge de la
branche à priori.
Remarques et premières conclusions
L’hypothèse du pipe model se vérifie à l’échelle des branches mais pas à l’échelle de
l’ensemble des rameaux. Il existe un effet position des branches sur les valeurs de η,
mais on n’observe pas de relations simples pour expliquer ces variations dans les
analyses présentées ici.
B. Évaluation du volume occupé par les feuilles en fonction de l’échelle de

représentation de la couronne
1. Introduction
Afin de pouvoir calculer simplement et rapidement un bilan radiatif sur des couronnes
d’arbres, les modèles de transfert radiatif définissent une représentation géométrique
simplifiée de la couronne. Dans Hi-sAFe, la couronne est représentée par un ellipsoïde
de révolution.
Cette simplification géométrique engendre probablement une erreur au niveau de

l’estimation du volume de la couronne, ce qui se répercute sur le calcul du rayonnement
intercepté par celle-ci. Dans un premier temps, nous ne travaillerons que sur le volume
de la couronne.
Une des manières d’évaluer cette erreur est de suivre l’évolution du volume de la
couronne en fonction de l’échelle de simplification de celle-ci. C’est ce qui a été effectué
sur deux noyers hybrides âgés de 7, 8 et 9 ans et sur un noyer âgé de 9 ans.
2. Matériels et Méthodes
Le volume occupé par le feuillage a été évalué par le modèle ellipsoïdal. Chaque
ellipsoïde est calculé à partir des axes d’inertie du nuage de feuilles considérées. Le
centre de l’ellipsoïde est l’isobarycentre du nuage de feuilles. Les axes de l’ellipsoïde
caractérisent donc l’orientation générale de la forme.
L’implémentation de ces algorithmes dans AMAPmod a été très récemment effectuée

par F. Boudon (2004) lors de sa thèse, en collaboration avec C. Godin et C. Pradal. Ces
outils sont actuellement en cours de développement.
Les trois noyers présentés précédemment ont été utilisés. Les maquettes des noyers
3311 et 3610 permettent de suivre l’évolution des caractéristiques de la couronne à 7, 8
et 9 ans. Le noyer 162 n’est représenté qu’à 9 ans.
Le volume du feuillage a été évalué à trois échelles : la couronne (C), les branches (B)
et les pousses annuelles (PA) (figure 6). Le volume de feuillage à l’échelle C est
approché par le volume de l’ellipsoïde (figure 6b) ; le volume de feuillage à l’échelle B et
PA est approché par la somme du volumes des ellipsoïdes à l’échelle considérée (figure
6 c et d).

INRA Report
a b c d
Table 90 Représentation de la couronne du noyer 3311 à 9 ans. Représentation

des feuilles uniquement (a) et des ellipsoïdes à l’échelle de la couronne (b), des
branches (c) et des pousses annuelles (d). Les feuilles vertes sont portées par des
pousses longues et les feuilles bleues par des pousses courtes. Chaque carré au
sol représente une surface de 1m².
Soit (Va)i le volume de feuillage de l’ellipsoïde i calculé à l’échelle a et (Vb)j le volume
de feuillage de l’ellipsoïde j calculé à l’échelle b. Par la suite, on pose :
∑ (Va)
i
i
δ (a/b) =
∑ (Vb)
j
j
Si δ(a/b) > 1, il existe un espace vide (non occupé par le feuillage) à l’échelle a.
Inversement, si δ(a/b) < 1, les volumes à l’échelle b se recoupent. En revanche, si
δ(a/b) ≅ 1, cela ne signifie pas forcement qu’il n’existe pas d’espace vide dans a car les
volumes à l’échelle b peuvent se recouper.
Ce ratio a été calculé pour représenter de manière synthétique le facteur multiplicatif du

volume occupé par le feuillage lorsque l’on passe d’une échelle à une autre.
3. Premiers résultats et discussion
Les figures 7 et 8 représentent l’évolution du volume de feuillage en fonction de l'échelle

macroscopique de représentation de la couronne et de l'âge du noyer (3311 et 3610
respectivement).

INRA Report
20
couronne branches pousses
18
16
14
Volume (m3)
12
10
8
6
4
2
0
7 8 9
Âge du noyer
Table 91 Évolution du volume de feuillage (m3) en fonction de l'échelle

macroscopique de représentation de la couronne et de l'âge du noyer. Calcul du
volume avec le modèle ellipsoïdal. Noyer 3311.
20
18 couronne branches pousses
16
14
Volume (m3)
12
10
8
6
4
2
0
7 8 9
Âge du noyer

macroscopique de représentation de la couronne et de l'âge du noyer. Calcul du
volume avec le modèle ellipsoïdal. Noyer 3610.
45
couronne branches pousses
40
35
Volume (m3)
30
25
20
15
10
0
7
Échelle de représentation

macroscopique de représentation de la couronne. Calcul du volume avec le
modèle ellipsoïdal. Noyer 162
Quelque soit l’échelle, le volume total de feuillage augmente avec l’âge de l’arbre
(figures 7 et 8). Sur le noyer 3311, le volume de la couronne calculé à l’échelle C, B et

INRA Report
PA augmente d’un facteur 4.4, 4.0 et 2.5 respectivement entre 7 et 9 ans. En revanche,
ce même facteur calculé sur le noyer 3610 est de 3.2, 4.5 et 6.5.
La différence entre les deux individus à l’échelle PA est importante (2.5 et 6.5). Elle
s’explique sans doute par une différence du ratio pousses longues/pousses courtes
entre ces deux individus. En effet, le noyer 3610 à 9 ans est composé d’un grand
nombre de pousses courtes (tableau 1) du fait de la floraison à 7 ans. Or le volume
occupé par le feuillage à l’échelle des PA est égal à la somme du volume des ellipsoïdes
englobant les feuilles. Le nombre de pousses courtes étant important, il existe un grand
nombre de petits ellipsoïdes englobant les pousses courtes avec chacun un volume ε
(égale au volume de « vide » laissé entre les feuilles et l’ellipse). Si ε est indépendant de
la taille de l’ellipse, alors le volume calculé à l’échelle des PA augmente artificiellement
en sommant le volume des ellipsoïdes.
La figure 10 représente l’évolution du ratio δ.
2.5
d(B/C) 3311 d(PA/B) 3311
d(B/C) 3610 d(PA/B) 3610
d(B/C) 162 d(PA/B) 162
2.0
ratio δ
1.5
1.0
0.5
7 8 9
Âge du noyer
Table 94 Évolution du ratio δ en fonction de l'échelle macroscopique de représentation de
la couronne (et de l’âge des arbres pour les noyers 3311 et 3610).Le ratio δ suit une
tendance opposée pour les deux individus 3311 et 3610. Pour le noyer 3311, δ
augmente avec l'âge, ce qui suggère qu’au cours de la croissance, l’espace non occupé
par le feuillage augmente à l’échelle de la couronne ou des branches. En revanche, le
résultat contraire observé sur le noyer 3610 suggère que l’espace non occupé par le
feuillage diminue. Ce résultat peut peut-être s’expliquer par l’augmentation du nombre
de pousses courtes (suite à la floraison en 2001) qui comble les espaces libres avec les
rosettes de feuilles portées par les pousses courtes. Ces résultats doivent donc être pris
avec précaution et demande confirmation et discussion. Il suggère également le besoin
de définir un indicateur tel δ que j’ai défini rapidement, qui puisse prendre en
considération l’agencement des entités entre-elles. Ce travail pourrait être complémenté
en utilisant différents modèles de simplification pour évaluer le volume occupé par le
feuillage.
C. Calcul du biovolume ligneux des noyers 3311, 3610 et 162
1. Objectif
Dans le modèle Hi-sAFe, l’allocation du carbone dans la couronne est un phénomène

paramétré par le rapport entre le biovolume de la bille de pied et le biovolume ligneux de

INRA Report
l'arbre (Vb/Va). Ce ratio est difficile à calculer sur le terrain et implique souvent des
mesures destructrices.
L’objectif de ce travail était simplement de déterminer ce ratio en utilisant les maquettes

architecturales des noyers 3311, 3610 et 162 pour lesquels la longueur et le diamètre
des pousses annuelles avaient été mesurés.
2. Résultats
Les résultas sont indiqués dans le tableau 2. Les maquettes utilisées sont représentées
dans la figure 11.
Les résultats que la bille représente 24 à 31% du biovolume ligneux total de l’arbre.
Tableau 2 : Ratio entre le biovolume de la bille de pied et le biovolume ligneux des

noyers 3311, 3610 et 162 (Vb/Va), et volume ligneux total des individus (exprimé
en m3). Les arbres sont âgés de 9 ans.
Noyers hybrides Vb/Va Volume ligneux

total (m3)
3311 0.31 0.04
3610 0.27 0.03
162 0.24 0.10
Table 95 : Représentation du squelette ligneux des trois maquettes

architecturales des noyers 3311 (a), 3610 (b) et 162 (c) âgés de 9 ans. La ligne en
pointillés représente la limite entre la bille et le houppier utilisé dans le calcul du
ration Vb/Va.

INRA Report
ReferencesBarthélémy, D. (2003). Botanical background for plant architecture analysis and modeling.
Plant growth modeling and application - Proceedings - PMA03. B. G. Hu and M. Jaeger. Beijing, Chine,
Tsinghua University Press, Springer: 1-20.
Berninger, F., M. Menccucini, et al. (1995). "Evaporative demand determines branchiness in Scots pine."
Oecologia 102: 164-168.
Berninger, F. and E. Nikinmaa (1997). "Implications of varying pipe model relationships on Scots Pine
growth in different climates." Functional Ecology 11: 146-156.
Boudon, F. (2004). Représentation géométriques multi-échelles de l'architecture des plantes, Université

Montpellier II: 176p.
Godin, C., E. Costes, et al. (1997). "Exploring plant topological structure with the AMAPmod software : an
outline." Silva Fennica 31: 355-366.
Godin, C., E. Costes, et al. (1999). "A method for descritibing plant architecture which integrates topology
and geometry." Annals of Botany 84: 343-357.
Shinozaki, K., K. Yoda, et al. (1964). "A quantitative analysis of plant form - the pipe model theory."
Japanese Journal of Ecology 14: 97-105.
Sievänen, R., E. Nikinmaa, et al. (2000). "Components of functional-structural tree models." Annals of Forest
Science 57: 399-412.

INRA Report
3 Contractor 1: INRA – UAFP

Contractor 1 : INRA – UAFP (FRANCE)

UMR DYNAFOR, BP 27,
31326 Castanet Tolosan cedex, France
(Unité Agroforesterie et Forêt Paysanne belongs to UMR DYNAFOR since 1st january 2003 (same
address)

Name Unit Tél Fax E-mail

Dr Cabanettes A. DYNAFOR 33 5 61 28 53 82 33 5 61 28 54 11 cabanett@toulouse.inra.fr
Mr Gavaland A. DYNAFOR 33 5 61 28 54 19 33 5 61 28 54 11 gavaland@toulouse.inra.fr
Technical assistance
Laurent Burnel, Jérôme Willm and Adrien Montupet
Time spent on the different workpackages during the third year
Name WP1 WP2 WP3 WP4 WP5 WP6 WP7 WP8 WP9 WP10 Total
Cabanettes A. 1.30 1.30
Gavaland A. 3.37 0.55 3.92
Total 4.67 0.55 5.22
TECHNICAL ASSISTANCE : Laurent Burnel, Jérôme Willm and Adrien Montupet
TIME SPENT ON THE DIFFERENT WORKPACKAGES (man.month)
Cabanettes A. 1.1 1.1
Gavaland A. 1.0 1.0
Total 2.1 2.1
WP3: Silvoarable experimental network (2.1 persons-month)
“Report of field measurements and analysis of the first results”
During the 6 last months of the SAFE project, research activities of DYNAFOR team
consisted in:
-achieving root description of two black walnut trees at Les Eduts (one in a silvo-arable
field and one in a forest stand) as mentioned in the previous report

INRA Report
-measuring intercrop rape samples from Grazac previously harvested (166-167/2004)
-harvesting and measuring intercrop maize samples at Pamiers
-analyzing the above experimental data.
All these activities and the main related results are presented below.
Moreover, one article 4 submitted on 23th july 2004 has been accepted with minor
revisions and another one 5 has been submitted during autumn.
1
Final submission of the first article will be performed before the deadline: 16th march
2005.
4
CHIFFLOT V., BERTONI G., CABANETTES A. and GAVALAND A., 2004. Beneficial effects of
intercropping on the growth and nitrogen status of wild cherry and hybrid walnut trees. Submitted
to Agroforestry systems 23th july.
5
LAMBS L., MULLER E., CHIFFLOT V. and GAVALAND A., 2004. Sap flow measurements of
wild cherry trees (Prunus avium) in an agroforestry system during a dry summer, South-west of
France. Submitted to Annals of Forest Science 15th November 2004.

INRA Report
Root description at Les Eduts
Four black walnut trees have been felled in the beginning of 2002 at Les Eduts for stem
analysis and root description.
Stem analysis has been performed immediately after felling but root description has
been delayed for time availability reason. Root description is very time consuming so it
has been decided to do it on only two trees, one in a silvo-arable field (AF1, field “Foix-
Bonnet”) and one in a forest stand (F1, “La Foye).
Two different methods have been used for the description of AF1 and F1 root systems.
Root description of AF1 tree (110-114/2004)
methods
The root system has been divided in elementary segments considered as homogeneous
for direction.
For each segment several parameters have been registered:
-distance to the trunk centre of the beginning of the segment
-depth to ground surface of the two ends of the segment
-diameter of the two ends of the segment
-azimuth of the segment
Each root has been removed after entire descriptions (of all its segments) in order to
avoid confusion between roots.
Unfortunately, it has not been possible to describe all the roots because of the difficulty
to clear them: the soil was very stony and many roots were very long and reached the
middle (or more) of the cultivated alley occupied at the moment by winter wheat.
results
With this operation, it has been possible to describe 163 root segments of a total length
of 70.70 m and a volume (calculated using equation (1) 6of 0.242 m-3. The average
diameters of the segments were 46 mm (beginning, range: 180-9) and 35 mm (end,
range: 126-2).
With the recorded data, it has not been possible yet to calculate the xyz position of the
measured segments in order to study the spatial repartition of the roots.
Root description of F1 tree (320-323/2004)
6
equation (1) Volume = 1/3*PI*h*(R²+R*r+r²)

INRA Report
methods
A 3D digitizer (Picture 1) connected to a computer has been used:
By clicking at the beginning and at the end of each elementary segment of the root
system, it has been possible to record 3D position of the segment; at the same time,
diameter at the two ends of the segment has been measured with a calliper and
recorded on the computer.
Picture 1: 3D digitizer next to root system of F1 tree – Les Eduts
After description, the overall root system of F1 tree has been weighed (roots have been
weighed separately form stump and taproot); 10 samples have been kept for Dry Matter
(%) and density (kg.m-3) estimation.
The 3D data have been analyzed by Dr F. Danjon (INRA Bordeaux-Pierroton) with

AMAPmod software.
results
270 roots (269 + stump and taproot) have been observed through the description of
1290 root segments of an average length of 14 cm. The main parameters of the root
system (from F. Danjon calculations) are presented in Table 96.
Table 96: Main parameters of the F1 tree root system – Les Eduts
Total root length (m) 177
Total root volume (m3) 0.129
Stump and taproot volume (m3) 0.014
Total root volume without stump & tap root (m3) 0.105
Maximum radial distance from the trunk (m) 4.69
Maximum depth of roots (m) 1.08

INRA Report
The described root segments were from roots of 2nd to 6th order (1st order is stump and
taproot). The average diameters of 2nd to 6th order roots were respectively 55.4, 26.2,
18.3, 12.5, and 0.9 mm.
Table 97 gives the distribution in number, volume and length of the root systems
between the 2nd to the 6th order level. It clearly shows that the main part of the root
system volume is in the 2nd and 3rd order roots; at the opposite, the 2nd order roots
represent a small part in length of the root system comparatively to 3rd to 5th order roots.
Table 97: Repartition of the F1 root system according to root order

(without stump and taproot) – Les Eduts
From digitizing, 3D views of the root system of F1 tree have been obtained. The
comparison between a photograph (Picture 2) and the 3D view with the same orientation
(Picture 3) suggests that it is possible to get rather faithful representations with the
digitizing technique.

INRA Report
Picture 2: Root system of F1 tree Picture 3: Root system of F1 tree

(photograph) – Les Eduts (from digitizing) 7 - Les Eduts
(Picture F. Danjon)
It is also possible to analyze the root distribution according to depth and orientation. First
observations (Picture 4) suggest that the root distribution of trees in a forest stand such
as F1 trees are not dependant on the row direction. It may depend whether distances
between two next trees in the row and between the rows are the same (7m in our
situation) or not.
The repartition of the root biomass in length, volume and number among 8 directions
(that is to say eight 45° angles) is presented in Table 98. It appears that about one third
of the root system is in the West direction; this is in agreement with picture 4 which
shows that a long and big root is in this West direction,
An explanation of this distribution could be that dominant wind and rainfall come from the
West, this should imply for the trees to have more available water on this side and to
develop roots for stability?
7
The different order roots are represented by different colours

INRA Report
4
Picture 4: Vertical view of the root system of F1 tree (from digitizing) – Les Eduts
(Picture F. Danjon)
Table 98: Root distribution of F1 tree according to 8 directions – Les Eduts
%
35
30 % number
% volume
25 % length
20
15
10
0
North North- East South- South South- West North-
East East West West

INRA Report
DM ratio obtained from the 10 oven-dried root samples is 62.9 %.
The volume of these samples has been estimated with equation (1) and a 448 kgDM.m-3
ratio has been obtained.
Total weight of the observed root system was 58.9 kg DM (37.9 % from stump and
taproot and 62.1 % from the roots).
Discussion
The repartition of weight between stump and taproot and the other roots of F1 tree is not
in agreement with the volume repartition obtained from digitization: stump and taproot
represent 37.9 % of the total root system weight and only 18.5 % of the volume. It is
likely that the stump and taproot volume has been under-estimated by digitalization
because of the difficulty to reach the lower parts and to click at the right position.
Nevertheless, another stump and taproot volume estimate can be obtained using its
weight and the 448 kg DM.m-3 ratio presented above.
The overall root system volume estimate of F1 tree could thus be:
0.105 + 0.050 (instead of 0.024) = 0.150 m3.
A part of the root system has not been described and the AF1 and F1 tree root system
volumes have probably been underestimated. The underestimation is low for F1 tree
because more time has been spent for clearing roots of F1 before description than for
AF1 tree and also because F1 tree rooting system was much smaller (about twice
smaller) and easier to be described.
Nevertheless, the results can be linked with the estimates of the above-ground biomass
estimation (presented in 2004 report). Table 99 gathers the results for above-ground and
below-ground biomass of AF1 and F1 trees.
Table 99: Biomass repartition between above-ground and below-ground

compartments of F1 and AF1 trees – Les Eduts
F1 AF1
DMKg m3 % total DMKg m3 % total
Total 103.28 63.7 276.56 NA
Bole 34.46 21.3 86.24 NA
Above-ground Crown 68.62 42.4 190.32 NA
biomass Total 0.223 59.8 0.561 69.9
Bole 0.083 22.3 0.176 21.9
Crown 0.140 37.5 0.385 48.0
Below-ground DM 58.92 36.3 NA NA
biomass Volume 0.150 40.2 0.242 30.1
DM 162.20 100 NA NA
Total tree biomass
Volume 0.373 100 0.803 100
It appears that either in forest stands or in silvo-arable stands, the part of biomass in the
bole could approximately be the same (21-22 %) but that the crown biomass could be

INRA Report
less important in the first situation (37-42 % in F1 tree) than in the second one (48 % in
AF1 tree); at the opposite, % belowground biomass should be more important in forest
stands (36-40 % in F1 tree) than in silvo-arable stands (30 % in AF1 tree). Nevertheless,
these first conclusions need to be confirmed by further investigation and to be compared
with available results in the scientific literature.
Concerning the spatial distribution of tree roots, using the digitizing technique in silvo-
arable plots would give information on the effect of intercropping on the tree root
development. Unfortunately, it is necessary to clear roots previously and this needs a lot
of time especially if the soil is stony like at Les Eduts.

INRA Report
Intercrop yield measurement at Grazac
The yield measurement has been performed with the advises of Mr Garric from CETIOM
8
.
Methods
Rape (variety Toccata) has been settled at Grazac during autumn 2003 (295/2003).
Plants countings have been performed twice during the crop season, 309/2003 and
69/2004, and also just before harvest (166/2004).
Rape has been harvested on 181/2004 but before, samples have been collected two
weeks (166 and 167/2004) before the whole harvest in order to estimate rape yield.
Samples for plant counting and samples for yield estimation where the same.
At Grazac, intercrop can be located in three kinds of locations:
- near wild cherry rows
- near hybrid walnut rows
- in crop alleys without trees.
For each of these kinds of locations, three gradients have been chosen for yield
measurements (one in each replication); each gradient consisted in six 1 m² square
sample (three at the east and three at the west of a tree (for plots with trees), the centre
of the sample squares number 3 and 4, 2 and 5, 1 and 6 being respectively at 1.5 m,
2.75m or 4m distance from the tree as presented in Table 100.
Consequently 54 (6 x 3 x 3) gradient samples have been observed.
The average DBH and height of the trees in the centre of the gradients were (8.4 cm,
504 cm) and (7.5 cm, 427 cm) respectively for wild cherry and hybrid walnut trees
(measurements 323-331/2003); concerning wild cherries, the three gradients have been
chosen in order to be near the same clone: Monteil.
Moreover, nine 1 m² “average square samples” (3 kinds of location x 3 replications) have

been harvested, these average samples being the further as possible from wild cherries,
or walnuts, or in the centre of the pure crop plot alleys (Table 100).
8
CETIOM: Centre Technique Interprofessionnel des Oléagineux Métropolitains
http://www.cetiom.fr/CTMSite/page/cetiom/english/01_growers_f.htm

INRA Report
Table 100: Rape samples location in the intercropped alleys - Grazac

West East
sample number : 6 5 4 3 2 1
0,25m 1m
1m "gradient square sample"
tree row
6 metres
tree row
"average square sample"
10 metres
Table 101 gives the map of Grazac field experiment with the 63 (54 + 9) rape samples
location within the field.
Table 101: Rape samples location at Grazac
The harvest has been performed on 166-167/2004 (Picture 5 and Picture 6).

INRA Report
Picture 5: Rape harvest at Grazac Picture 6: Rape harvest at Grazac

(Stems have been cut with pruning (Harvested plants have been put in
shears) bags)
After harvest, rape samples have been let at the air during two months under a
greenhouse with bags open for drying (Picture 7).
Then, threshing has been performed with an electric thresher (Picture 8) from 219/2004
to 230/2004.
Picture 7: Drying rape samples under a Picture 8: Electric thresher used for
tunnel greenhouse – INRA Toulouse threshing
rape samples (INRA Toulouse)
The suction fan of the thresher has not been used during the threshing in order to avoid
grain loss because of the very light grain weight of rape. Grain samples have been
cleaned afterwards with a sifter and with a small electric fan to blow the bigger impurities

INRA Report
For the 1000 grain weight measurement, grains have been counted with a Numigral
(Durant solid State 1800).
Thus, from the harvested rape samples, to sub-samples have been obtained after
threshing and cleaning: straw (including impurities) and grains.
The sub samples have been oven-dried: twice at 80°C during 48 h for straw and only
once at 110°C during 72 h for grains. The 1000 grains samples have been oven-dried at
50°C during 72 h.
Weighing of samples has been performed with electronic scales with a 10-1 g precision.
So, for each (1 m²) sample, 4 parameters have been recorded:
1- Number of plants, 2- DM grains, 3-DM straw, and 4- 1000 grains weight.
Results
Plant counting
There were neither significant effect of the sample location (near wild cherry, near
walnut, pure crop plot) nor of the gradient number (distance to the tree) on the number of
plants.
The results obtained were [mean (standard-deviation)]:
- On 309/2003: 25.2 (9.8) plants.m-2
- On 69/2004: 23.8 (9.8) plants.m-2
- On 166/2004: 23.4 (9.9) plants.m-2, that is to say 234000 plants.ha-1.
Rape yield
The results are presented in Table 102.
The average yield observed at Grazac was rather low:
5.2 tons DM.ha-1 above ground biomass (1.06 ton grain + 4.14 tons straw)
The yields were also very heterogeneous (standard deviation of grain yield is 0.68
ton.ha-1) due to heterogeneous soil fertility: the density (plants.m-2) of rape was not
significantly different between samples.
There were no significant differences between the 9 “average square samples” (far from
the trees and in the middle of the alleys, Table 100).
Among the 54 “gradient samples”, there was a significant effect of the gradient location,
near trees (wild cherry or hybrid walnut) or in the pure crop plots, but no effect of the
location in the gradient (1 to 6), on DM grains weight.

INRA Report
There was also a significant effect of the location in the gradient (1 to 6) on the 1000
grains weight, but only for the gradients near trees (not for the pure crop gradients):
1000 grains weight at locations next to the tree (3 and 4) was significantly lower than
yields at locations more next to the alley centre (1, 2, 5 and 6).
No significant effect was observed on DM straw (and total above ground DM biomass).
Table 102: Rape yield parameters at Grazac in 2004

Number Number of DM grains DM straw Above- 1000 grains
of plants (g.m²) (g.m²) ground DM weight
Kind of samples samples (ha-1) (g.m²) (g)
All 63 23.4 106.0 414.3 520.3 3.71
Average 9 32 114.3 461.4 575.7 3.76
Gradient : all 54 21.9 104.6 406.5 511.1 3.70
Wild cherry (A) 18 23.8 a 94.5 b 406.1 a 500.6 a 3.67 a
Hyb. walnut (B) 18 22.2 a 90.0 b 380.1 a 390.1 a 3.76 a
Pure crop 18 19.4 a 129.2 a 433.3 a 562.5 a 3.67 a
Gradient(A)+ (B) 36 22.9 92.3 393.1 485.4 3.71
East (1+2) 12 24.6 a 84.8 a 361.8 a 446.6 a 3.78 a
Near tree (3+4) 12 20.3 a 95.8 a 420.9 a 516.8 a 3.47 b
West (5+6) 12 23.9 a 96.2 a 396.6 a 492.8 a 3.88 a
Comparisons are made vertically. Means followed by the same letter are not significantly
different (5%).
As a conclusion, it can be said that the main effect of the trees on the crop development
at Grazac during the crop season 2003-2004 could have been to reduce the filling of
rape grains during the maturation period because of the shading effect of trees (between
10 and 20 % of radiation are kept by trees as presented 2004 report); this effect does
not seem to be strong enough at this stage of tree development to induce significant
yield loss : the main yield differences observed at Grazac in 2004 were due to soil
heterogeneity.
Further investigation in more homogeneous situation needs to be performed!

INRA Report
Intercrop yield measurement at Pamiers
Methods
Maize has been sown on 122/2004 with variety ‘Bounty’.
The soil has been fertilized twice:
-before sowing, 115/2004 (35 N, 70 P, 70 K)
-after sprouting, 147/2004 (161 N)
Chemical weeding has been performed the day after sowing (123/2004), with Primeset
(5 liters.ha-1) + Lagon (0.6 l.ha-1).
Maize has been irrigated twice in July before flowering.
The entire harvest has been performed on 301/2004 but samples of above-ground
biomass (grains + straw) have been collected previously (285/2004) for yield estimation.
Each sample was a 2 meters long part of maize row.
Two kinds of samples have been collected:
-“average samples” located in the centre of the alleys and at the cross of the diagonals
between four trees, in a location visually observed as locally optimal for crop yield. 12
average samples have thus been harvested.
-“gradient samples” located on both sides of a wild cherry in a line perpendicular to the
tree row. On this line, every two meters part of maize row has been collected. 7
gradients of 12 samples each have been harvested; among the 7 gradients, 2 were
control ones without tree.
Table 103 gives the location of these samples in the wild cherry plantation at Pamiers.

INRA Report
Table 103: Location of the 2004 maize samples in the wild cherry plantation of
Pamiers
North
Location of the gradients Maize

Location of the average samples
For each sample, the entire above-ground biomass has been harvested; the ears have
been separated from the rest of the plant and shelled manually.
The samples have been oven-dried (during 48 hours at 80 °C); then several yield
parameters have been measured:
-grain weight
-biomass yield (total above ground biomass without grain)
-1000 grains weight (measured on 500 grains sub-samples counted with a Numigral
counter)
Yield parameters variation of maize according to the distance to the tree row has been
analysed using the “supsmu” function of Splus software package (Table 106 to Table
110).
Results
Wild cherry measurement

INRA Report
The wild cherries in the centre of the gradients had an average DBH of 24.9 ± 2.6 9 cm
and an average crown surface of 30.6 ± 5.9 m² (average crown radius of 3.12 ± 0.3 m) 10;
the average crown height was 5.64 ± 0.36 m.
Spacing
The average distance between maize rows was 75 ± 7.4 cm and the average distance
between the tree row and the next maize row was 155 ± 40 cm; the average density of was
87400 plants.ha-1 (spacing 75 cm × 15.25 cm).
Considering the low variation of spacing between maize rows, it has been decided to
apply the mean, 75 cm to every measurement unit for the yield calculations
(measurement unit area of 1.5 m² (2m × 0.75m)).
Maize yield
Table 104 gives the average results in the different situations observed in the
experiment.
Table 104: Maize yields according to the kind of sample at Pamiers in 2004
Kind of Number DM % DM 1000 Grains DM % DM Above- Harvest
sample of grains grains grains number straw straw ground Index 11
samples weight by ear DM
(ton.ha- (ton.ha- (%)
1 1
) (g) ) (ton.ha-
1
)
Average 12 7.02 77.8 224 383 5.56 70.92 12.58 54.7

sample
± 2.89 ± 2.5 ± 37.5 ± 133 ± 2.08 ± 13.39 ± 4.75 ± 7.1
(41.1) (3.3) (16.7) (34.6) (37.4) (18.9) (37.8) (13.0)
Gradient 59 4.09 74.5 191 282 3.99 72.4 8.08 44.4

with
trees ± 3.15 ± 15.1 ± 43.3 ± 115 ± 2.15 ± 8.8 ± 5.20 ± 17.7
(77.2) (20.2) (22.8) (40.6) (53.8) (12.1) (64.3) (39.8)
Gradient 23 9.63 77.4 291 407 7.35 60.77 16.98 56.3

without
trees ± 3.49 ± 2.0 ± 21.5 ± 111 ± 2.48 ± 6.7 ± 5.84 ± 4.78
9
mean ± standard deviation
10
Crown measurements have been performed on 289/2004
11
Harvest Index = DM grain / Above-ground DM Biomass

INRA Report
(36.3) (2.6) (7.4) (27.2) (33.4) (11.1) (34.4) (8.5)
mean ± standard-deviation (Variation Ratio = standard deviation / mean, %)
The first data analysis (Table 104) has led as to conclude to a very great heterogeneity
of the yields through the field, partly due to the presence (or not) of trees but mainly due
to soil fertility heterogeneity: Even for samples without trees, the variation ratio (VR) of
yields (DM grain and DM straw) are very high, between 33 and 36 %; this ratio is
somewhat higher for “average samples” (37 to 41 %), probably because of some tree
effect, and much higher in the “gradient samples with trees (between 64 and 77 %).
DM % of grain are less varying (VR is near 3 %), except for gradients with trees (20 %).
DM % of straw is more heterogeneous than DM % of grain (11 to 19 %).
In order to take into account this heterogeneity and to go further in the data analysis,
every sample yield parameter in a gradient has been compared to the maximum
obtained in the gradient. Table 105 gives the average yields observed in the gradients
with (A) and without (O), in % of the maximum yield in the gradient) and an estimate of
the yield reduction with trees: grain yield reduction is about 24 % while straw yield
(above-ground biomass production without grain) is not much affected by the presence
of trees (only 6 % reduction).
Table 105: Yield parameters (average of the % of the maximum in the gradient) -
Pamiers
Kind of sample DM grains Grains DM straw Above-ground DM
number by
ear
Gradient with trees (A) 52.64 66.33 60.97 57.87
Gradient without 69.28 71.31 64.80 68.28
Trees (O)
Loss with trees
100 ×(1 - A/O) 24.02 6.98 5.91 15.25
Yield parameters (in % of the maximum in the gradient as explained above) have been
plotted with the distance to the tree in Table 106 (grain yield) and Table 107 (Harvest
Index).

INRA Report
Table 106: DM grain yield (% of the maximum yield in the gradient) according to
the distance to the gradient centre - Pamiers
From Table 106 it can be concluded that grain yield is reduced when the maize row is at
less than about 3.50 m of the tree row; the reduction reaches half of the average yield of
the gradient without trees (0.5 × 69.28 %, Table 105) at less than 2.5 m from the tree
row. In the measurement units next to the tree row (less than 2 m) grain yield is very low.
Table 107: Harvest Index according to the distance to the gradient centre -
Pamiers
Table 107 shows that the Harvest Index (HI) is also influenced by the distance to the
tree row in the same way as grain yield. Average HI is 56 % in the gradients without

INRA Report
trees and 44 % in the gradients without trees (Table 104); this HI reduction is mainly
located in the maize rows next to the tree rows (less than 3 m).
The 1000 grains DM is higher in the gradients without trees (291 g) than in the gradients
with trees (191 g) but in the latter ones, the variation according to the distance to the tree
row does not appear clearly (Table 108).
Table 108: 1000 grains DM weight according to the distance to the gradient centre
- Pamiers
If we consider the grain number by ear, the average difference between gradients with
trees and gradients without trees is not very important (Table 105) but this parameter is
influenced by the distance to the tree row mainly for maize rows next to tree rows, 3 m or
less (Table 109).

INRA Report
Table 109: Grain number by ear (% of the maximum in the gradient)

according to the distance to the tree row - Pamiers
Discussion
The common question for all the yield parameters presented above is whether or not,
the tree effect is the same on both sides of the tree, north and south. Table 110 shows
that grain yield variation is not so different between north and south sides of a tree,
except may be for the higher distances. But, the south of a tree is the north of the tree in
the next row!
As a consequence, a common model for both sides of trees linking yield parameters and
distance between maize row and tree row could be fitted. An interesting observation is
that for a distance of 3.12 m (the average crown radius of the trees), the yield could be
50 % of the potential yield (maximum in the gradient).
A Weibull model [Y = a × (1-exp(-b × xc] with parameters (a, b, c) dependant on

dendrometric parameters, crown radius or DBH ?, and spacing between tree rows
seems to be convenient in this situation…

INRA Report
Table 110: Grain yield in gradients with trees (% of the maximum yield in the
gradient) - Pamiers
50 % ?
Crown limit
Our data provide information on the tree effect in the north – south direction, but
information is missing concerning the tree effect on maize yield in the east-west
direction. The “average samples”, compared with results in the further maize rows in the
gradient were supposed to give this missing information; unfortunately, maize yield at
Pamiers is too much heterogeneous. Gradients with measurement units from parts of
maize rows of 6 meters long divided in three parts (2 meters long each) should have
been a more appropriate method for evaluating the effects of wild cherries on the maize
production.

INRA Report
Abstract of the Toulouse reports from 2002 to 2005

The main contribution of UAFP to SAFE project consisted in providing experimental data
within WP 3. Four silvo-arable sites have been studied, three field experiment previously
settled by UAFP (Lézat, Grazac and Pamiers) and one silvoarable farm (Les Eduts).
The first site, Lézat, settled in december 1997, has only been studied (for tree growth)
during the first year of the project because i) the main forest species, Sorbus domestica,
was not a major species in the SAFE project ii) the landowner died in 2002 and his farm
was proposed for sale.
In the second site, Grazac, also settled in December 1997, two forest species, wild
cherry and hybrid walnut trees have been planted respectively at 10m × 6 m and 10 m ×
10 m spacing. Three kinds of management of 8 m out of the 10 m wide alleys have been
compared i) intercropping ii) chemical weeding iii) fallow (no management). All the trees
have been measured annually; moreover, mineral nutrition of trees has been
investigated during three years, 2001 to 2003, and tree phenology has been described
during the growing season 2003; some investigations on water supply of trees (water
potential, tensiometry and sapflow) have also been carried out during year 2003.
Intercrop yield measurements have been performed during years 2000 (wheat), 2001
(wheat), and 2004 (rape).
The site of Pamiers consisted in a wild cherry clone field trial previously settled by INRA
in spring 1986. Three different tree spacing are being compared in this site (11m ×11 m,
11m ×5.5m and 5.5m ×5.5m). The alleys between tree rows were occupied by
spontaneous (ligneous and herbaceous) vegetation. During spring 2002, spontaneous
vegetation and deep ploughing has been performed in the 11 m wide alleys and
intercrops have been managed in half of the 11 m wide alleys during the growing
seasons 2002 (maize), 2003 (wheat) and 2004 (maize), the other half part of the 11 m
wide alleys has been kept as “fallow control” and the rest of the field trial is from now on
considered as “forest control”.
Annual measurements of DBH has been carried out before the first intercrop settlement
and then after the growing seasons 2002 and 2003. Moreover, maize yield
measurements have been carried out in autumn 2002 and 2004.
At Les Eduts, height and DBH measurements have been performed during year 2003 in
one of the black walnut stands of Mr Claude Jollet in which trees were either in forest
condition, 7m × 7 m without intercrops, or in silvo-arable conditions, 14 m ×7m with
intercrops managed in 12 m of the 14 m wide alleys since the settlement of trees (almost
30 years). Then four black walnuts, two in silvo-arable fields and two in a forest stand,
have been felled for stem analysis (on the 4 trees) and root description (on two of the
four trees, one in silvo-arable field and one in forest stand).
Moreover, hemispherical photographs have been carried out at Grazac, Pamiers and
Les Eduts during year 2003 (and some ones in the beginning of year 2004). These 1476
photographs have been analyzed with GLA and Surfer software to get the evolution of
the radiation transmitted by the tree canopy during the growing season and the variation
of this parameter according to canopy size and tree species.
Among the amount of experimental data collected during the SAFE project, the main
emerging conclusion is that trees grow much better when there are associated with

INRA Report
intercrops. This result has been observed in all of our four sites. When intercropping is
performed since the tree settlement, significant differences on height and diameter tree
growth are obtained since the third year. A positive effect of intercropping on diameter
increment has also been recorded after two years of intercropping in a wild cherry stand
of 18 years. The result is impressive when intercropping is performed during many
decades: intercropped black walnuts grown at Les Eduts on poor soils have produced
three times more above-ground ligneous biomass, and at least twice more below-ground
biomass, during 25 years than black walnuts in forest stands. This gain in overall
ligneous biomass results in an important gain in valuable timber: bole volume is
increased by +132 % after 25 years. The effect of intercropping on heartwood formation
dynamics has to be investigated more further.
The consequence is that, the rotation length of trees could be reduced significantly
with intercropping, of may be 15 to 20 years at Les Eduts.
The main explanation of this better tree growth could be that trees benefit from
intercrop fertilization and get better nutritional status: they produce more foliar
biomass and the nutrient uptake from the soil is increased: mineral concentrations,
mainly nitrogen, of leaves are increased with intercropping.
Intercropping could be more profitable to some tree species, probably the more
demanding ones: at Grazac, it appears that the gain in growth increment is much more
important for hybrid walnut than for wild cherry trees.
The second conclusion is that trees reduce intercrop yields: a part of the radiation is
kept by the tree canopy and is not available for the intercrops; the yield reduction will
thus be much more important for summer crops than for winter crops.
The yield reduction is very low when the trees are young: a significant yield reduction
has not been clearly demonstrated at Grazac with 7 years old wild cherry and hybrid
walnut trees; this effect is more obvious with older (and larger) trees: at Pamiers, the
average maize yield reduction was near 25 % with 19 years old wild cherries with Dbh of
25 cm planted at 11 m distance between rows; the reduction could reach 50 % at the
limit of the tree canopy and more than 50 % under the tree canopy.
Nevertheless, the experience of Mr Claude Jollet shows that intercropping can be

carried out during more than two decades without important yield reduction: at Les
Eduts, intercropping has been carried out during more than 25 years between black
walnut tree rows. After 25 years, the intercropped alley (with winter crops) was still 12 m
wide with 14 m spacing between tree rows.

INRA Report
4 Sub-contractor to INRA :ICRAF

ICRAF FINAL 6-MONTHS REPORT
Sub-contractor to INRA: ICRAF
Sub-contractor 1: ICRAF (Indonesia)
ICRAF – World Agroforestry Centre, Jl. CIFOR, Situ Gede, Bogor, 16610, Indonesia.
Principal Investigators:
Name Tel E-mail
Meine van Noordwijk + 62 251 625415 m.van-noordwijk@cgiar.org
Betha Lusiana + 62 251 625415 b.lusiana@cgiar.org
Contribution to Work Packages
WP5 Belowground Interactions
In the last 6-month period ICRAF focuses on producing the algorithm for water and N
uptake. The final products that available are:
1. Water uptake algorithm: as a document and its instanteneous implementation in

excel worksheet
2. N uptake algorithm: as a document and its instanteneous implementation in excel

worksheet
3. Three draft manuscripts:
A process-based algorithm for sharing nutrient and water uptake between

plants rooted in the same volume of soil I. Water in static root systems
by
Meine van Noordwijk, Betha Lusiana, Christian Dupraz, Simone Radersma, Harry Ozier-
Lafontaine, and Peter de Willigen
A process-based algorithm for sharing nutrient and water uptake between plants
rooted in the same volume of soil II. Nutrients in static root systems
by
Betha Lusiana, Meine van Noordwijk, Christian Dupraz and Peter de Willigen

INRA Report
A process-based algorithm for sharing nutrient and water uptake between plants
rooted in the same volume of soil III. Growing root systems
by
Meine van Noordwijk, Betha Lusiana, Rachmat Mulia and Christian Dupraz
All the above products are attached.
Sensitivity Analysis
Sensitivity analysis on water and N uptake algorithm (2D version) as implemented in

WaNuLCAS model was conducted.
In this study, the WaNuLCAS model was used to simulate a Peltophorum dasyrrachis –
maize intercropping system. The intercropping system is based on experiments in
Lampung-Indonesia and parameterized by Suprayogo (2002), with soil of 10% clay and
10% silt. Mean annual rainfall in Lampung is 3100 mm. Simulations were made for a
second cropping season (March – May), after the hedgerows have established for 6
months. The effects that variation of tree and crop root length densities have on the
uptake and partitioning of water were simulated. Tree roots are assumed to be constant
over the whole simulation while crop roots are dynamic as a function of growth stage.
Figure 1. Effects of root length densities on water and N uptake by tree and crop in
Peltophorum – maize agroforestry system, as simulated in WaNuLCAS. Water and N
use by crop, tree and both plants in agroforestry system is expressed relative to its value
at default root length density. Water and N use by crop or tree in monoculture system is
expressed relative to its value in agroforestry system at default root length densitiy..
Both water and N uptake shows similar trend (Figure 1). Decreasing root length density
for the tree will have only a small effect on calculated water and N uptake (over 85
days) in a monoculture of the trees, but it will lead to a small increase in crop water
uptake and decrease in tree water uptake in a mixed tree + crop system, for near-
constant total water uptake. Decreasing crop root length density will decrease total
water and N uptake of the mixed system as well as in a crop monoculture, but will have
a moderate positive effect on tree water and N use and strong negative effect on crop
water and N uptake in the mixture system.
Reference
Suprayogo D, Hairiah K, van Noordwijk M and Cadisch G. 2002. The inherent ‘safety-
net’ of Ultisols: measuring and modelling retarded leaching of mineral nitrogen.
European J. Soil Science 53, 1-10

INRA Report
5 Sub-contractor to INRA :CTL

CONTRAT EUROPEEN n° QLK5-CT-2001-00560 (Code INRA B3645)
SILVOARABLE AGROFORESTY FOR EUROPE (SAFE)
Contrat de Sous-Traitance INRA / AGRO.M - Centre de Transfert
Rapport final de l’activité du technicien
Décembre 2004
Projet de sous-traitance signé le 16/11/01 s’étalant sur une période de 3 années (clôture le
15/10/04) suivant les clauses du contrat INRA B.36.45.
I . Relevé des coûts par tranche (journées travaillées pour SAFE) :
BILAN DES DEPENSES REALISEES (EUROS HT)
Coût
Poste 1ère tranche 2ème tranche 3ème tranche Coût final
prévisionnel
4,5 HM 5,3 HM 4 HM 13,8 HM 13,5 HM

Personnel (Ingénieur,
Techniciens, M.O.O)
9 074,10 € 11 003 € 8472 € 28549,1 € 25957 €
Frais divers de gestion 725,90 € 1 980 € 678 € 3383,9 € 2077 €
TOTAL 9 800 € 12 980 € 9150 € 31930 € 28034 €
(HM : Homme Mois)
L’activité de sous-traitance représente 13,8 hommes mois sur la période des 3 années du
programme pour une coût total de 31930 euros HT. L'engagement initial portait sur 12
HM et un budget de 28034 euros. Le solde a versé en fin de contrat est de 5 504 euros.
II . Réalisation des tâches (article 1 du contrat) :
Travail à effectuer Objectif réalisé
Mesure du développement architectural aérien

Oui
et souterrain des arbres et des cultures.
Mesure du bilan radiatif et hydrique des

Oui
associations arbre/culture.

INRA Report
Mise en forme et interprétation des données

Oui
obtenues pour le modèle de simulation.
Les résultats sont remis sous forme informatique (base de données) de façon conforme au
cahier des charges établi pour l’élaboration du modèle Hi-sAFe, logiciel de modélisation
agroforestier, développé par les autres partenaires du projet. (Fichier disponibles non
joins à ce rapport)
III . Compte rendu détaillé de l’activité :
1 . Installation de sites expérimentaux :

a . Macroclimat local :
Installation ou remise en état des stations météo des sites expérimentaux de

Restinclières (34) et Vézénobres (30).
Programmation et enregistrement au pas de temps horaire des variables instantanées

suivantes : Tair, Hrair sous abri. Rayonnement (RPA et/ou global). Vent moyen horaire.
Pluie cumulée horaire.
b. Mouvements de la nappe alluviale :
pose de piézomètres en témoin agroforestier et forestier à 4 ou 5 mètres de profondeur

suivant le terrain.
nivellement des piézomètres installés.
pose des mires limnimétriques dans le Lez.
c. Teneurs en eau du sol
installation de tubes de sonde à neutron pour la mesure de la teneur en eau du sol. Les
tubes sont installés à 3 mètres de profondeur dans deux transects pour l’agroforestier :
arbre/culture et arbre/arbre.
d. Dynamique saisonnière de croissance des arbres
Installation de dendromètres à ressort pour le suivit fin de la variation et la croissance en

diamètre des arbres (parcelle agroforestière de Vézénobres (30)).
2 . Suivis et mesure de sites expérimentaux :

a . Macroclimat local :
relevé bimensuel des données.
b. Suivis de la hauteur de la nappe alluviale :
mesure au pas de temps bimensuel de la profondeur de la nappe

INRA Report
mesure de la hauteur d’eau dans le Lez en amont et en aval
c. Mesure de l’homogénéité des pratiques culturales
uniformité de l’épandage d’azote
homogénéité des infestations d’adventices
d. Teneurs en eau du sol
mesures mensuelles des teneurs en eau du sol sur 3 m de profondeur à la sonde
e. Densité de racine fine des arbres et leur phénologie
Prélèvement séquentielle (1 tous les 2 mois de mars à novembre, soit 5 prélèvements)

de carottes de 1 à 2 m de profondeur pour suivi de la densité de racines fines des
arbres.
Après carottage : remplissage du trou avec de la terre fine, complété après une semaine
pour compenser le tassement.
f. Dynamique saisonnière de croissance des arbres
Mesure mensuelle du diamètre des troncs au millimètre pour valider le module de

croissance (allocation du carbone) de l’arbre.
Divers :
Soutien technique ponctuel aux divers protocoles de SAFE.
Travaux annexes (achat de matériel, entretien général du matériel ...).

Wageningen University Report
6 Contractor 2: Wageningen University

Contractor 2 : Wageningen University (WU, The Netherlands)

Department: Plant Sciences, Group Crop and Weed Ecology (CWE)
Agrotechnology and Food Sciences, Systems and Control Group (SCG)
Wageningen University, 6700 AK Wageningen, PO Box 430, Netherlands

Principal Investigators
Name. Tel Fax E-mail
WU
W. van der Werf 31-317-484765 31-317-485572 Wopke.vanderwerf@wur.nl
H. van Keulen 31 317 475955 31-317-484892 Herman.vankeulen@wur.nl
K. Keesman 31-317-483780 31-317-483780 Karel.Keesman@wur.nl
K. Metselaar 31-317-485972 31-317-485972 Klaas.metselaar@wur.nl
GPG
J. van den Briel 31-317-453 462 31-317-412740 gpg@het.nl
A. Kofferman 31-317-453 462 31-317-412740 gpg@het.nl
V.G.F. Repelaar 31- 553-781 239 31-553-781 284 deelerwoud@tiscali.nl
FINIS
F. Schuman 49-404-14314909 49-404-14314909 f.schuman@agropark.de
B. Schindler 49-4545 789151 49-4545 789151 schindler@finis-ev.de
During the reporting period (1 August 2004 – 31 January 2005) no EU-paid additional
staff was employed by Wageningen University (EU budget was not sufficient to cover
salary costs of Klaas Metselaar until the end of his contract on 5 August 2004). Most
work at this contractor was done by permanent personnel, notably K. Keesman, W. van
der Werf & H. van Keulen, in interaction with – especially - the partners 1: INRA
(Christian Dupraz & Martina Mayus), 5: Cranfield University (Anil Graves & Paul
Burgess) and 8: FAL (Joao Palma & Felix Herzog).
The first main objective of WU during this period was to develop the simple agroforestry
model Yield-sAFe into an instrument for yield forecasts under practical conditions
(Workpackage 6B). The development of Yield-sAFe was taken up by WU when it
became clear that the comprehensive detailed model Hi-sAFe would not be suitable for
rotation-long simulations. Developing Yield-sAFe for practically useful yield forecasts
required in depth analysis of model response to different calibration approaches, which
was conducted in collaboration with Cranfield University, FAL and INRA during the
reporting period. Contacts were maintained through email, phone calls and working
visits. The objective of calibrating Yield-sAFe to practical conditions was challenging,
notably because very few calibration data for Yield-sAFe were available (see contractor
5 report), and because the model had to be calibrated for widely diverging environmental
conditions. The second main objective of WU was to publicize the scientific outcomes of
its contribution to SAFE.
Coordination of subcontractors in the frame of WPs 2, 7 and 9 was maintained by Mrs dr

Martina Mayus, who was employed by INRA until 31 November 2004.

The main achievements of WU are described in the report of work package 6B and
include:
♦ A preliminary report on time series of tree and crop yield in agroforestry systems
(Klaas Metselaar, August 2004) and uncertainty analysis of the agroforestry
model Yield-sAFe (Klaas Metselaar, August 2004).
♦ Frequent and in depth contributions to the calibration efforts with Yield-sAFe by

partners 5 and 8. An intensive session on model balibration issues was
conducted during the final SAFE meeting, 2-5 November 2004 in Zurich. A
successful working visit to Cranfield was made by Karel Keesman and Wopke
van der Werf on 2 and 3 December 2004 to help overcome critical difficulties in
model calibration (report made by Anil Graves).
♦ W. van der Werf supported M. Mayus in efforts to conduct a sensitivity analysis

of Yield-sAFe to be included in a scientific paper describing the model concept of
Yield-sAFe (van der Werf et al. 2005).
♦ A contribution was made to the preparation of the national conference in the

Netherlands (to be held on 20 April 2005).
♦ At the final SAFE meeting in Zurich (2-5 November 2004), W. van der Werf, K.
Keesman & H. van Keulen participated actively in critical discussion on model
choice for yield forecasts. Issues with respect to scientific plubaction of SAFE
results were discussed.
♦ K. Keesman prepared in collaboration with W. van der Werf & H. van Keulen an
in depth mathematical analysis of a parameterised version of Yield-sAFe, which
was then submitted to the Bulletin of Mathematical Biology (submitted December
2004).
♦ H. van Keulen, in collaboration with M. Mayus, supervised the MSc student

Michel Postma, who made an inventory of farmers’ opinions on Agroforestry in
the Netherlands (report provided to Fabien Liagre) and a literature study about
the potentials of Agroforestry in the Netherlands (MSc report Michel Postma).
Total workload over the reporting period was 1.0 man-month for K. Keesman, 0.75 man-
month for W. van der Werf, and 0.5 man-month for H. van Keulen.
Sub-contractor FINIS e.V.: contributions to WP9

Policy: Task 9.2
FINIS e.V. investigated the political and legal basic conditions of Agroforestry in
Germany. Since in Germany after the constitution of the Federal Republic the Lands
have the right of their own legislation, we examined additionally the responsible
guidelines of the agriculture, forestry and environmental protection in Schleswig-
Holstein, Baden-Württemberg and Hessen. The results were summarized in a detailed
report, which was provided to the WP9 leader Gerry Lawson.

Silvoarable plot: Task 9.4
During the year 2004, strong winds caused again and again damages to the tree
bamboo pole. Thus, labour intensive repair work was required after each strong storm,
to avoid that the tree could be damaged and impaired in its growth. It appeared that the
bamboo became brittle and got crannies by the weather. Finally, we replaced the
bamboo poles by wood poles, before autumn storms could cause large damages. The
new poles resisted well two strong autumn storms; future maintenance work after storms
should be small.
At the end of the growing period 2004, the height of all trees was measured for a second
time. We compared the results with the measurements from the previous year. It turned
out that the tree species had very differently developments. While Prunus avium and
Acer saccharum showed a clear increase in height, Acer pseudoplatanus mainly
invested in tree foliage, and only few in height growth. It must be noted, that Acer
pseudoplatanus grew better during the season 2004, than during the previous year.
From this we concluded, that the Acer pseudoplatanus first invests in root development
before it starts growing aboveground. Whether this assumption is true will show the
growing season of 2005.
In co-operation with FINIS, the farmer will maintain the silvoarable field also after the
SAFE project has been closed.
Further contributions
For the SAFE meeting in Zurich in November 2004, FINIS supported by M. Mayus
provided a Power Point presentation about the silvoarable field in Groß Zecher,
Schleswig-Holstein (Safe web site). The presentation, hold by M. Mayus, included, the
tree heights and climatic data from the region (2002 and 2003).
Every year in October, the broadcast NDR (Nord Deutscher Rundfunk, north German
broadcast) organizes a "day of the open farm". The public has the possibility of visiting
different farms in Schleswig-Holstein to experience agricultural production practices.
FINIS, used the opportunity to present agroforestry at the farm Groß Zecher. Next to an
exhibition of photos of various agroforestry systems, and a presentation of the SAFE-
project and a guided visit on the silvoarable field was undertaken. The number of visitors
was between 1000 to 1500.
In January 2005, Frank Schumann hold a lecture about Agroforestry at the Academy for
Science and Culture of the donation Herzogtum-Lauenburg in Mölln, Germany. At the
same day, there was the prelude the by FINIS organized exhibition on agroforestry
systems, which continued till 25 February 05.
Sub-contractor GPG: Contribution to WP9

GPG investigated the political and legal basic conditions of Agroforestry in the
Netherlands: The following authorities/ organisations had been contacted for information,
Alterra (research), Probos (forestry advice), Bosschap (forest policy institute for
regulation of the sector an ‘public-private’ organisation) and Bosgroepen (cooperation of
private forest owners). Alterra and the Bosschap could not answer the questions

because they were not sufficient informed about agroforestry. The subject is yet too
new.
The information obtained by Probos are:
1. What is the availability/capability of forestry subsidies for agro forestry?
For afforestation and forest management subsidy is possible in the frame of the subsidy
regulation called ‘Programma Beheer’, provided that no subsidy is obtained for other
purposes /management forms. Subsidies for combinations of agriculture and forestry
are not possible in principle (from forestry point of view approached). From agricultural
point of view however – we come closer – there are subsidies for the maintenance of
traditional (‘high stem’) fruit orchards. A pastoral combination is there the most common.
2. There are no reactions known about agro forestry
3. The definitions:
A. According to the Boswet (Forest Law): definition forest:
Where and for which counts the Boswet?
The Boswet is alone of application outside the town centre/ in rural areas. The municipal
council decides upon the areas/ limits of the zone. It needs for this approval from the
provincial authority (representative states (GS) of the province). The town centre
forestry law is not necessary to fall together with the town centre in the frame of the
traffic- and road law.
The Forest Law (is outside the towns) of application for:
1. Forests /woods, shelterbelts and windbreaks with a surface larger than 10 Ares
2. Trees planter in a row (linear plantations / windbreaks) of more than 20 trees.
The Boswet is not of application on:
- property and gardens;
- hedges of hawthorn (one row) if these are as such planted and managed;
- one rowed linear plantation of poplar or willows, on or along agricultural lands

and along roads;
- Italian poplar, Tilia tree, chestnut and weeping willow;
- Fruit tree;
- windbreaks along orchards;
- ‘Christmas trees’ plantations not older than twelve year, cultivated on special for
that purpose designated grounds;

- Seedlings etc. in nurseries.
B. Definition ‘high stem orchard’ (fruit) according to subsidy regulation agricultural

nature management (is part of the ‘Programma Beheer’, see above);
- being an orchard of apple-, pear-, plum-, cherry- or , walnut species;
- the orchard having a surface of minimally 25 Ares, with a density between 50

and
150 trees per hectare.
- The trees are at least 4 meters high.
For the management of a fruit orchard subsidy is possible when:
- it stands the criteria (see above);
- the orchards lays in two formally classified landscapes (‘hill country’ and ‘rivers
territory’);
- fulfils criteria for proper ‘natural’ management: no manure, no chemical pest

control etc.;
- In the cases of pear- or apple species when the trees are at least 1 time per two
year being trimmed/ pruned.
C. There are no definitions in use yet for ‘good (agricultural) practice’ (as
element/criteria for ‘cross compliance'. Most close comes the subsidy criteria for
orchards (see just above).
D. There is no solid (unique) definition of ‘agriculture’. ‘Agriculture’ means in the

practice, arable or grassland /meadow farming, not including waste land,
windbreaks, forest, nature etc. In other words; there is no unambiguous precise
definition.
4. The Dutch subsidy schema is not favourable for agroforestry. Trees are just not
seen as an element of agriculture.
From the Bosgroep, we got the next contribution:
“By my knowledge there are in the Netherlands no subsidy possibilities for

Agroforestry. The EU financial contribution for afforestation in the Netherlands is
subject to the condition that the agricultural land is taken out of production and is
legally converted in into (formally classified) forestland this means: it may no longer
be used for agricultural production. The road to subsidy for agroforestry seems to be
blocked by this.”
“As the definition for agroforestry is concerned they cannot help us any further.
Because agroforestry makes no part of subsidy schemes or sets of instruments no
definitions are necessary. Possibly that Freerk Wiersum of the University of
Wageningen may help you further. He is engaged in the subject agroforestry.

The Bosgroep does not consider orchards as a form of agroforestry, as to their opinion it
has in the Dutch context nothing to do with forestry.”
Silvoarable plot: Task 9.4
In spring 2004 most of the replanting was done. The trees of Picea abies, planted in
spring 2004, are growing well. In this report period, some Robinia trees has been
replanted. The farmer will maintain the silvoarable field also after the SAFE project has
been closed.
During the SAFE meeting in Zurich in November 2004, M. Mayus gave a presentation
about the Dutch silvoarable field and the policy situation in the Netherlands and
Germany.

NERC Report
7 Contractor 3: NERC - Centre for Ecology and

Hydrology
Contractor 3 (NERC – Centre for Ecology & Hydrology)

Centre for Ecology and Hydrology - Maclean Building, OX10 8BB, Wallingford, Oxfordshire, United
Kingdom


Dr. Nick Jackson +44 1491 692 336 +44 1491 692 424 naj@ceh.ac.uk
Dr John Roberts +44 1491 692 334 +44 1491 692 424 jro@ceh.ac.uk
Mr Gerry Lawson +44 1793 341 1925 +44 131 445 3943 gela@nerc.ac.uk
Dr Deena Mobbs +44 131 445 8559 +44 131 445 3943 dcmo@ceh.ac.uk
Dr Marcel van Oijen +44 131 445 8567 +44 131 445 3943 mvano@ceh.ac.uk
Dr Bob Bunce 44 153 953 2264 +44 153 953 4705 Bob.Bunce@wur.nl
Time Contributions
Year 1
Name WP1 WP2 WP3 WP4 WP5 WP6 WP7 WP8 WP9 Total
Dr. Nick Jackson 0.6 0.6 0.6 0.6 0.6 0.6 0.6 4.2
Dr John Roberts 0.2 0.2 0.2 0.2 0.2 0.2 0.2 1.4
Mr Gerry Lawson 0.4 0.4 0.4 0.4 0.4 0.4 0.4 2.8
Dr Deena Mobbs 0.3 0.3 0.3 0.3 0.3 0.3 0.3 2.1
Dr Bob Bunce 0
Total 1.5 1.5 1.5 1.5 1.5 1.5 1.5 10.5
Year 2
Dr. Nick Jackson 0 0 4.5 0 0 0 4.5
Dr John Roberts 0 0 0.15 0 0 0 0.15
Mr Gerry Lawson 0 0 0 0.5 0 0.65 1.15
Dr Deena Mobbs 0 0 0 0.35 0 0 0.35
Dr Marcel van Oijen 0 0 0 0.6 0 0 0.6
Dr Bob Bunce 0 0 0 0 1 0 2
Total 0 0 4.65 1.45 1 0.65 7.75
Year 3
Dr. Nick Jackson 0.0 0.0 2.6 0.0 0.0 0.0 2.6
Mr Gerry Lawson 0.0 0.0 0.0 0.0 0.0 2.7 2.7
Dr Bob Bunce 0.0 0.0 0.0 0.0 0.6 0.0 0.6
Total 0.0 0.0 2.6 0.4 0.6 2.7 6.3
Year 4 (6 months) WP1 WP2 WP3 WP4 WP5 WP6 WP7 WP8 WP9 Total
Dr. Nick Jackson 2.5
Mr Gerry Lawson 1.5
Dr Bob Bunce 0.5
Total 2.5 0.5 1.5 4.5
Overall Total 11.3 1.9 3.6 6.4 29.1
WP5 Below-ground interactions
NERC-CEH Wallingford (Nick Jackson) leads WP5, but almost all the work in this
workpackage was completed by month 36 (mainly year 2). Re-design and
implementation of the modelling by staff at both INRA-APC (Guadeloupe) and
INRASYSTEM (Montpellier) was largely completed by month 36, and no further working
visits to Guadeloupe were conducted. Nick Jackson attended the final workshop in

NERC Report
Zurich and the Consortium management Meeting. The final report on Workpackage 5 is
largely unchanged from the 36 months version.
WP8 Scaling up to the farm and region

CEH Edinburgh and Dr RGH Bunce (sub-contractor associated with both CEH and
Alterra in Wageningen) have provided advice, student supervision, and land-use
classifications for FAL (Partner 8). The European Land Classification system has been
completed and used to define landscape test sites in Spain, France and the
Netherlands. Dr Bunce attended the final SAFE workshop in Zurich and submitted a
final report titled ‘the distribution of silvo-arable systems in western Europe and their
ecological characteristics’ (Annex 1). This described contrasting samples in the Atlantic
European zone and in Southern Europe. Silvoarable systems were extremely rare in the
former, but in the latter silvo-pastoral and silvo-arable system covered millions of
hectares. It is very difficult to separate these two land-uses using existing methodology
since the land is often used for both purposes. There are likely to be significant
landscape ecological benefits to linked to the trees in new silvo-arable plantings,
especially in homogeneous lowland landscapes dominated by arable crops.
WP9 European guidelines for agroforestry

This Work Package is co-ordinated by GJ Lawson (NERC), was particularly active
during the final two months of the project.
Task 9.1 – Document the problems of farmers establishing new silvoarable plots
in at least 5 countries.
Anecdotal evidence was gathered during end-user meetings (between January and
March 05) in several countries on the perceptions of officials and farmers. In summary,
officials have difficulties with agroforestry for several reasons.
• There is no EU Forest Policy (or mention of forestry in the Constitution)

• There is little knowledge that knowledge that agroforestry is mentioned several
times in 1999 EU Forest ‘Strategy’
• There is lack of experience of old or new agroforestry systems, or willingness
to be flexible with grant rules in order to benefit experimental trials of
agroforestry.
• Agroforestry presents complications to calculating grant levels which are most
easily solved by making it ineligible.
• Agroforestry has complicated effects on the ‘cadastral’ and local tax status of
land.
• Responsibility for agroforestry falls between agriculture, forestry and
environment departments (the Agriculture Department wants to hang on to
agricultural land, the Forestry Department doesn’t believe its possible to grow
good quality timber at wide spacing, the Environment Department doesn’t like
regimented rows, intensive management and control of weeds).
• Finally, there is a perception that EU doesn’t allow it !! (e.g. it is frequently
stated that ‘EU insists that afforestation grants must reduce agricultural
surpluses’).

NERC Report
Farmers are reluctant to introduce agroforestry plantations because of technical

difficulties, such as:
• uncertainties over management, time consumption and yield;

• likely damage to field drains;
• perception of increased pest problems;
• incompatibility with machinery & potential tree-damage;
• little knowledge of timber markets;
• possible lower timber quality;
• trees owned by landlord and not tenants.
Or because of disincentives due to current regulations:
• low or no subsidies following 1257/99 (no or lower arable area payments, no

or pro-rata reduced planting grants, no income support payments, ineligible for
agri-environmental payments);
• classification as permanent forest land (lower tax but lower land value &
irreversible planning control);
• time and bureaucracy for grant application process;
• scepticism of professionals and advisors.
Task 9.2 – Compare eligibility of silvoarable systems for Government financial
Support in EU member states

A Report on Deliverable 9.2 was prepared during just after the reporting period (Annex
2). More detailed appendices have been prepared for the UK, France and Spain, and
are under production for other SAFE member countries. Rules and regulations in many
countries are changing and it is hoped that these country reports can be updated after
the end of the project. Results are presented in the Workpackage 9 report, and are
summarised in Table 9.1.
Presentations focusing on eligibility of agroforestry systems for current arable area

payments, for tree-planting grants, and for future single-farm-payments were made
following the formal end of the project at end-user meetings in three countries (Paris -
26th January 05, Madrid - 11th March 05, and Brussels - 30th March 05).

NERC Report
Table 9.1 Eligibility of Agroforestry Systems for Agricultural and Forestry Grants under current Pillar I and Pillar II rules
Country Agricultural Payments (First Pillar) Forestry Grants for Agroforestry Spacings (Second Pillar) Agri-environmental
(Second Pillar)
Arable Area Payment Livestock payments in Planting Tree-maintenance Income support (for
silvopastoral systems (if (usually for 5 8-20 years)
declared as ‘foreage years)
areas’).
France Yes on cropped area. Yes in grazed woodland if Yes, proportion of Yes, proportion of Yes, on non-cropped Yes, specific AF measure,
(with young trees) or area forage area >50% total cost. But strong total cost area. But not yet but only applied in two
reduced for crown area restrictions in practice applied for Departments
(mature trees) (such as follow up of
the plantation by a
research institute)
Germany Yes on cropped ares, but Yes in grazed woodlands if No No No Possible but untried
so far no references with forage area >50%
mature trrees
Greece Yes on cropped area Yes in grazed woodlands if No No No Possible but untried
reduced by crown area forage area >50%
Italy Yes usually reduced by Yes in grazed woodlands if No No No Possible but untried
crown area but can vary forage area >50%
Netherlands Yes on cropped ares, but Yes in grazed woodlands if No No No Possible but untried
so far no references with forage area >50%
mature trrees
Spain Arable payments usually Yes in grazed woodlands Proportion of total Yes, proportion of No Small grants in some
reduced by more than (e.g. Dehesas) if forage cost (density as low total cost Regions but only for
12
corwn area. area>50% as 278t/ha for some maintaining existing trees
species
Switzerland ? Yes in grazed woodlands No No No Possible if AF recognised as
Ecological Compensation
Areas
UK Yes on cropped area, Yes in grazed woodlands Pro-rata reduction Pro-rata reduction No Possible (e.g. hedges) but
provided connected to (tho’area is reduced to from 1200t/ha for untried
larger field. account for trees & grazing poplar (only). No
should be possible for 7 income support
months per year)
12
E.g. support for traditional agroforestry in Andalucia; maintenance of non-productive trees (Aragon, Madrid); maintenance of windbreaks and setos (Asturias, Canarias, Cataluna,
Rioja; Pais Vasco; soil protection through lines of trees and scattered trees (Pais Vasco);

NERC Report
Task 9.3 – Comment on proposed changes in forestry and agroforestry policy based
on scenario testing using models.
Partners in WP6, 7 and 8 have collaborated to develop plot and landscape models of
agroforestry growth which allow the effects of subsidies for tree, crop and environmental to
be varied. Standard scenarios are being tested using the FARMSAFE model within
Landscape Test Sites in Spain, Netherlands and France, and to a lesser extent in other
countries. These scenarios form part of the final milestone for the SAFE Project (Milestone
15), and have been described in the WP7 and WP8 final reports.
A final conclusion on the profitability of agroforestry within the reformed CAP depends on the
whether agroforestry is considered eligible for Single Farm Payments, and whether countries
implement Article 41 of the draft Rural Development Regulation 2007-20013. This
regulation, for the first time, provided for tree-planting grants to be paid for trees at
agroforestry spacings. The SAFE Project has identified 7 policy issues.
Regulation 1782/03 introducing the move to the ‘decoupled’ Single Payment Scheme (SPS)
indicates that ‘woods’ (Article 43) and ‘forests’ (Article 44) are ineligible for the SPS. But
confusion exists because the Regulation does not define either ‘woods’ or ‘forests’. Already
there are examples of farmers removing trees from farmland (e.g. traditional orchards in
England, hedges in Poland or dehesa systems in Spain) because they fear the loss of SPS
payments.
Guidance Document AGRI-2254-2003 recommends that the threshold of 'woodland' is > 50

stems per ha, but does allow countries to define exceptions in the case of ‘mixedcropping’.“
In accordance with Article 5(1)(a) of Regulation (EC) No 2419/2001, areas of trees –
particularly trees with a potential use only for wood production inside an agricultural parcel
with density of more than 50 trees/ha should, as a general rule, be considered as ineligible.
Exceptions may be envisaged for tree classes of mixed cropping such as for orchards and
for ecological/environmental reasons. Eventual exceptions must be defined beforehand by
the Member States.” We propose replacing ‘tree classes of mixed-cropping’ with
‘agroforestry systems’ and include a simple definition of agroforestry.
Farmers obtaining the Pillar I SPS are obliged to demonstrate that they maintain the farm in
‘Good Agricultural and Environmental Condition' (GAEC). Annex IV of Regulation 1782/03
gives one GAEC condition as ‘avoiding encroachment of unwanted vegetation on agricultural
land'. EU countries differ in their definition of GAEC but it should be clear at the EU level that
well managed Agroforestry Systems fulfils GAEC requirements.
The draft Rural Development Regulation includes support for new planting of agroforestry
(Article 41) but NOT the 5-year maintenance element received by conventional plantations.
However, good maintenance during the 5 first years of a low density tree stand are crucial for
the success of the plantation. Tree protection, weed control, and stem pruning are essential.
Existing Agroforestry systems can be managed to maximise environmental benefits. These

traditional management costs could be included as an option within the agri-environmental
measures proposed by the draft RDR.
Regulation 2237/03 Chapter 5 sets levels and conditions for subsidies to nut plantations. • It
sets minimum densities (125/ha for hazelnuts, 50/ha for almonds, 50/ha for walnuts, 50/ha
for pistachios, 30/ha for locust beans) but indicates that payments to nut trees orchards will
NOT be made if these are intercropped. • This condition is reflected in national legislation,
but is an unreasonable condition provided that SPS is not claimed.

NERC Report
The 1998 EU Forest Strategy emphasised Agroforestry in the context of: ‘sustainable and
multifunctional management of forests … including optimisation of agroforestry systems’
(p15); – research to concentrate on ‘… diversification (nonwood uses, agro-sylvo-pastoral
systems)’..(p16); – maintenance of traditional management of silvopastoral systems with high
levels of biodiversity which may be lost of these areas area abandoned (p23); – the
importance of agroforestry for carbon sequestration (p23) Yet agroforestry is hardly
mentioned in national forestry strategies, or current EU or national rural development
strategies, or in the recent publication on ‘Sustainable Forestry and the European Union’.
The SAFE project has produced 4 key policy proposals.
Proposal 1: Agroforestry systems refer to an agriculture land use system in which high-
stem trees are grown in combination with agricultural commodities on the same plot.
The tree component of agroforestry systems can be isolated trees, tree-hedges, and
low-density tree stands. An agroforestry plot is defined by two characteristics: a) at
least 50% of the area of the plot is in crop or pasture production, b) tree density is less
than 200/ha (of stemsgreater than 15 cm in diameter at 1.3 meter height), including
boundary trees.
Proposal 1: A definition of agroforestry is suggested that includes isolated trees, treehedges

and low-density tree stands, which clearly distinguishes between agroforestry and forestry.
Proposal 2: The total area of an agroforestry parcel should be
eligible for the Single Payment Scheme
Proposal 2: This proposal is compatible with existing Regulations, removes the

contradiction between the two pillars of the CAP on rural trees (farmers will no longer be
stimulated to remove trees to get CAP payments), and simplifies controls, and therefore
saves a lot of European money
Proposal 3: Agroforestry systems should be backed by the Rural

Development Regulation (RDR, CAP second pillar)
Proposal 3: The draft RDR for 2008-2013 includes a welcome and innovative Article 41 that
introduces support for the establishment of new agroforestry systems. It could be
supplemented: a) to include maintenance costs for agroforestry planting in the same way as
in Article 40 for forest plantations; b) to support the eligibility of existing agroforestry systems
for improvement and environmental payments.
Proposal 4: The EU Action Plan for Sustainable Forest Management (2006)

should emphasise the need to maintain or increase the presence of scattered
trees in farmed landscapes (agroforestry)

NERC Report
Proposal 4: The 1998 EU Forest Strategy refers to agroforestry several times, but it was not
mentioned in the Commissions recent review of implementation of the Strategy. This
omission could be corrected in: a) the proposed Action Plan for Sustainable Forest
Management (2006), b) The EU Rural Development Policy Document (2006).
Task 9.4 – Co-ordinate the establishment by user participants in 3 countries of

silvoarable plots as a ‘social experiment’.
Replanting of dead trees and maintenance of crops have continued in the ‘social agroforestry
plantings’ in Netherlands, Germany and Greece. These are described in the WP9 report,
and were presented at the final SAFE workshop in Zurich in November 2004.
Significant difficulties during the reporting period.

Synthesis of recent changes in grants and subsidies and their impact on agroforestry
(Deliverable 9.2) were produced later than expected. This is partially because many
countries are still defining their interpretation of rules for the Single Payment Scheme. It is
hoped that these synthesis can be continued beyond the end of the SAFE project, partially
through the activities of end-user-groups in each of the member countries.
Results of the SAFE project will be presented at the annual meeting of the Farm Woodland
Forum (with which all UK members SAFE project are associated) on 29/6-1/7 at the
University of Wales Conference Centre at Gregynog, Powys, Wales. The theme of this
workshop is ‘Ecosystem Services of Farm Trees’, and the SAFE presentation will focus on
EU agri-environmental payments.

University of Leeds Report
8 Contractor 4: University of Leeds

2. Scientific team and time spent on the WPs
Name Telephone Fax E-mail
Dr D J Pilbeam +44 113 343 2895 +44 113 343 2835 d.j.pilbeam@leeds.ac.uk
Dr L D Incoll +44 113 343 2874 +44 113 343 2835 l.d.incoll@leeds.ac.uk
Dr F Agostini +44 1937 834 155 +44 113 343 2835 francesco.agostini@adas.co.uk
Ms F Reynolds +44 1937 834 155 +44 113 343 2835 f.h.reynolds@leeds.ac.uk
Mr C Wright +44 113 343 2858 +44 113 343 2835 c.wright@leeds.ac.uk
Dr M P Eichhorn m.p.eichhorn@leeds.ac.uk
David Pilbeam is leader of Participant 4, and is also coordinator of WP3. He is a senior

lecturer at the University of Leeds. Lynton Incoll has retired from a senior position at the
University (Principal Research Officer) during the course of the SAFE project, but has still
contributed to the project during the year. Francesco Agostini was a postdoctoral research
fellow employed full time on the SAFE project from April 2002 until December 2003. This
employment finished earlier than contracted, due to Dr Agostini moving to new employment.
Because of the strengthening of the Euro against stirling during the course of the SAFE
project funds allowed for a replacement for Dr Agostini, and instead of employing a research
fellow for the three months remaining on Dr Agostini’s contract it was possible to give a
contract for five months to allow for the replacement to have extra time to become familiar
with the project. The replacement was Markus Eichhorn, who commenced work on 24th May
2004. Dr Eichhorn was employed full time on the project until 12th November 2004.
Experimental Officer Fiona Reynolds did not work on the project during the six months, but
Experimental Officer Chris Wright, based at the School of Biology’s Field Research Unit,
worked on the project during that time.
1. Partner number, name and address of the participating organisation

Partner 4, UNIVLEEDS, UK
Address: School of Biology, University of Leeds, Woodhouse Lane, Leeds LS2 9JT,
UK.
3. Time spent on the different workpackages during year 4 (months).
0
D Pilbeam 0.20 0.20 0.40
L Incoll 0.50 0.05 0.55
F Agostini 3.75 0.22 0.22
F Reynolds 3.75
C Wright
M Eichhorn
Total 4.45 0.47 4.92

4. Contribution to workpackages
4.1 WP1 Silvoarable modelling strategies
Completed in year one.
4.2 WP2 European silvoarable knowledge

The WP2 objectives in the Technical Annex are:
O2.1 To collate the information gained in earlier studies including EU projects that
would be relevant to silvoarable systems and could fill gaps in the research of SAFE.
O2.2 To collect historic data from existing systems required for validating the plot-
scale yield model and economic evaluation.
A survey of modern silvoarable agroforestry systems and a literature search on

extinct systems were carried out in year one (Milestone 2.1, see first annual report). The
Leeds participants contributed to this by providing details of such systems in the UK. They
subsequently organised the rewriting of the report into a more consistent style, and with
additional material in year 3. In the current reporting period further amendments, particularly
in adding more quantitative information, were carried out. A large part of this work involved
adding maps showing the locations of sites of silvoarable agroforestry in Europe, so that
anyone looking at the report on the website would be able to see where precisely each site
is. Dr Incoll carried this out for all the UK sites included in the database. The maps used
were originally part of the Michelin series, but these did not give sufficient resolution to show
how to access each site, so they were replaced with maps from Ordnance Survey. This was
completed in month 42. Dr Eichhorn also added further references to the database of
publications on silvoarable agroforestry in the temperate zone that was largely completed in
year 3.
The largest part of the time spent during this reporting period was in taking the information in
the survey of extant silvoarable agroforestry systems in Europe and converting it into a
suitable format for publication in a scientific journal. This included adding further quantitative
information on the extent of existing silvoarable agroforestry systems in Europe and
illustrations. The finished paper was submitted for publication to the journal ‘Agroforestry
Systems’ in November 2004.
4.2.1. Results and deliverables
Improvement to the report on current silvoarable systems in Europe (delivered originally in

year one , improvements in place in month 36, further modifications month 42).
Provision of a database of publications on silvoarable agroforestry in the temperate zone

(month 36, some additions month 39).
A scientific paper on existing and historical silvoarable agroforestry practices in Europe was
submitted to the Journal ‘Agroforestry Systems’ in month 40. This was based on
contributions by Eichhorn,, M. P., Paris, P., Herzog, F., Incoll, L. D., Liagre, F., Mantzanas,
K., Mayus, M., Moreno, G., Papanastasis, V. P., Pilbeam, D. J., Pisanelli, A. and Dupraz, C.
4.3 WP3 Silvoarable experimental network

UNIVLEEDS (David Pilbeam) leads this workpackage. David Pilbeam and Chris Wright were
active in this workpackage during the reporting period.

During earlier reporting periods we:
Managed field experiments at our experimental site, carrying out measurements of tree
growth, crop growth, meteorological conditions according to agreed protocols.
Collected data on leaf fall and hemispherical photography from the Leeds experimental site
for model parameterisation.
Finalised the structure of the SAFE database.
Processed data from the organisers of the experimental sites, checked it and forwarded it in
an agreed format to P1 for inclusion in the SAFE database.
4.3.1. T3.1: Collect data from existing experiments as required by the

modelling activity
a) Working towards managing field experiments in a sound and concerted way with unified
protocols for field measurements
See Year 1 and Year 2 reports.
b) Working towards providing data from field experiments in a standardised format for model
parameterisation and testing
See Year 1 and Year 2 reports.
c) Collection of data from the Leeds experimental site
During the time of the SAFE project the Leeds site was managed with crops standard to a
cereal rotation in the UK. In the 2002/3 cycle oilseed rape was grown as a break crop, in the
2003/4 cycle winter wheat was grown. Details of the cultivation were given in the 36 month
report, but harvest of the crop did not occur until into the current reporting period. Tree
growth in the 2003 summer season was measured in March 2004, and for the 2004 summer
season will be measured in February 2005.
The 2003/4 winter wheat crop was harvested on August 2004. The 2004 harvest period in
the UK was affected by high rainfall, and harvest was extremely difficult. Some of the ears
had germinating grains present, and heavy rain had beaten down the crop. Harvest was
achieved by making one pass up each alley with a plot combine, and weighing the total grain
harvested over a set distance. Differences between the two control plots were high, because
of the amount of lodging of the crop, so that the control yield with which to make a
comparison is not particularly accurate.
It can be seen from Table 1 that yields of the crops were barely affected by the presence of
the trees for the first four years after planting. Indeed, for two of the first four years yields of
crops were actually slightly higher under the trees than in the control areas. After this time
there were significantly lower yields of crops under the trees (by ANOVA, P=0.05), except for
in 1999 (when yields were not significantly less). By 2003 yields were probably much lower
in the alleys, although the harvested crop was not weighed as pod shatter of oilseed rape
makes it a very difficult crop to obtain meaningful results for. The 2004 harvest was
inaccurate, due to the harvest difficulties referred to above, but appeared to show a reduction
in crop yields under the trees to less than 50% what was obtained in the control areas. This
is below the threshold at which a farmer would continue to grow crops in a commercial
silvoarable agroforestry system.

Table 1. Crop yields at the Leeds Experimental Site from 1992 to 2004. (Sowing and
harvest dates given as day of year and year).
Year of Crop Sole Silvoarable Ratio of

harvest crop yield (t ha-1 silvoarable
yield (t cropped to control
ha-1) area) yield
1992 Spring 6.34 6.62 1.04
barley
1993 Peas 5.46 4.83 0.88
1994 Winter 8.67 9.24 1.07
wheat
1995 Winter 8.17 7.81 0.96
wheat
1996 Winter 7.68 6.92 0.90
barley
1997 Spring 4.17 3.56 0.85
mustard
1998 Winter 10.55 9.55 0.91
wheat
1999 Winter 5.63 5.50 0.98
barley
2000 Winter 6.55 6.04 0.92
wheat
2001 Winter 6.38 4.70 0.74
wheat
2002 Winter 7.86 5.39 0.69
barley
2003 Winter * * *
oilseed
rape
2004 Winter 7.37 3.10 0.42
wheat
* not harvested due to pod shatter
The growth of the trees can be seen in Table 2a (for trees grown alongside alleys that were
continuously cropped) and in Table 2b (for trees grown alongside alleys that were
continuously fallow during the course of the experiment).
It can be seen that by 2004 the growth of the trees was noticeably higher alongside the
continuously fallow alleys, with timber volume at that time being 32% higher. The trees
adjacent to fallow alleys were taller (significantly so from 1994, by ANOVA, P=0.05) and had
greater diameter than trees adjacent to alleys that have been continuously cropped since
planting (significant by ANOVA, p=0.05, from 1995). This was due to a much larger
increment of growth in the trees next to the fallow alleys in summer 1995. For every season
since that time the increase in timber volume relative to the volume at the start of the season
has been slightly higher for the trees next to cropped alleys than for the trees next to the
fallow alleys.
In order to assess the productivity of the silvoarable plots it is possible to calculate Land
Equivalent Ratios. This is a common calculation in experiments on intercropping two annual
crops, but within the SAFE consortium has been controversial as it has not been used
extensively before to compare yields of mixed annual and perennial crops. If LER is
calculated as an annual value (LER =(crop yield per intercropped hectare/ yield of sole crop
per hectare) + (timber increment per season of intercropped trees per hectare/ timber
increment per season of trees by fallow alleys per hectare)) the results for the Leeds
Experimental Site are as shown in Table 3.

It can be seen that LER values are higher than 1.0 throughout the time of cropping, reaching
a maximum of 1.58 for the harvest in 2000. This indicates that the productivity of timber and
annual crops together is higher than the production of either would be on their own.
Table 2a. Leeds tree data: height, diameter at breast height and estimated timber
volume of four poplar hybrids in the continuously-cropped arable treatment at the
Leeds experimental site at Bramham from planting on 27th March 1992 to 31st March
2004. (Values are means, n = 60)
Day Year of Height Diameter at Calculated Estimated Estimated

of measurement (m) breast height cylindrical form factor timber volume
year (cm) volume (m3) (m3/tree)
87 1992 Estimate
1.20
356 1992 1.59
25 1994 2.52 2.36 0.001 0.42 0.000
25 1995 3.47 4.07 0.005 0.42 0.002
339 1995 4.46 6.15 0.013 0.42 0.006
3 1997 5.81 8.69 0.034 0.42 0.015
20 1998 7.50 12.00 0.085 0.42 0.036
40 1999 8.78 15.58 0.167 0.42 0.070
48 2000 10.11 16.63 0.220 0.42 0.091
79 2001 11.45 18.65 0.313 0.41 0.129
80 2002 12.81 20.40 0.419 0.41 0.172
72 2003 14.17 22.08 0.543 0.41 0.221
91 2004 15.21 23.97 0.690 0.40 0.280
Form Factor calculated from J M Christie, Yield Models for Forest Management, HMSO,
London, 1981.
Table 2b. Leeds tree data: height, diameter at breast height and estimated timber
volume of four poplar hybrids in the continuously-fallow arable treatment at the Leeds
experimental site at Bramham from planting on 27th March 1992 to 31st March 2004.
(Values are means, n = 60).
Day Year of Height Diameter at Calculated Estimated Estimated

of measurement (m) breast height cylindrical form factor timber volume
year (cm) volume (m3) (m3/tree)
87 1992 Estimate
1.20
356 1992 1.72
25 1994 2.64 2.70 0.002 0.42 0.001
25 1995 3.79 4.96 0.007 0.42 0.003
339 1995 5.44 8.77 0.033 0.42 0.014
3 1997 6.80 12.03 0.077 0.42 0.033
20 1998 8.36 15.18 0.151 0.42 0.063
40 1999 9.70 19.05 0.277 0.41 0.115
48 2000 11.00 20.36 0.358 0.41 0.148
79 2001 12.20 22.19 0.472 0.41 0.193
80 2002 13.72 23.89 0.615 0.40 0.249
72 2003 15.15 25.68 0.785 0.40 0.314
91 2004 15.92 27.26 0.930 0.40 0.370

Table 3. Annual values of LER (based on annual crop yield and annual increment in
timber volume) for the Leeds Experimental Site.
Year 1995 1996 1997 1998 1999 2000 2001 2002

LER 1.13 1.19 1.38 1.38 1.42 1.58 1.36 1.30
The LER values indicate that this is an efficient production system, which would be of benefit
both in terms of carbon sequestration per land area and in terms of maximising production
per land area so allowing farm land to revert to natural ecosystems. However, it should be
pointed out that the Leeds system is an experimental system set up for scientific
measurements. It would never be established for commercial agriculture as the tree spacings
have been set out to give a tree density similar to those found in farm woodland blocks of
poplar, and would not be suitable for agronomic operations on a commercial farm.
The Leeds site is part of the UK silvoarable network, three sites with the same experimental
design and growing the same crops each year. One of the other sites in the network,
Cranfield University, is part of the SAFE consortium. During the 2003/4 cycle both the other
sites discontinued cropping, leaving the Leeds site as the only one with a crop growing in
2003/4. This was because that work was funded from a different grant, which reached its
conclusion early in 2003. The measurements of tree height and diameter that will be made
in winter 2005 at Leeds will give data for fourteen years of tree growth with continuous
cropping, and will also give a comparison with trees grown beside continuously fallow alleys
over that time. No crop has been planted in the current winter period, and the alleys will be
put into set-aside in spring 2005. Tree growth will continue to be measured annually until
felling in 10-15 years time.
d) Implementation of database of consortium experiments
See year 3 report.
4.3.2 At the SAFE experimental sites specific information needed to

parameterise the biophysical model will be collected
See year 3 report.
4.3.3. Results and deliverables

The database of consortium experiments was compiled and posted on the SAFE website in
month 28 (D3.2).
An introduction to this database has also been posted, and it comprises the Handbook to the
European Experimental Resource (D3.1).
Measurements of tree phenology, tree heights and tree diameters at breast height have been
made at the Leeds site. Hemispherical photography of the trees after leaf fall was carried out
in year 3.
4.3.4. Future work

During the course of the SAFE project it was noticeable that the aspect of the Leeds
experimental plots was gradually changing from purely agricultural to more woody as the
trees matured. The plots are within 400 metres of a small area of woodland in which the tree
sparrow Passer montanus, a bird species of farm and woodland, is present. This is a
species that has shown a large population decline in the UK since the 1960s, with the
number of territories now estimated to be down to 110,000. It is on the Red list. Because
the experimental plots appeared to be developing into a suitable habitat for this species 9

nesting boxes were erected on trees in the plots not used for measurements. In the 2004
summer season three of these boxes were occupied by Passer montanus breeding pairs. It
is intended that bird nesting boxes will continue to be put out on the trees until they reach
maturity in 10-15 years time.
A paper on ‘Data Management for Decision Support Systems in Agroforesty’ is still being
prepared.
The Leeds group will contribute to the UK final SAFE conference in June 2005.
5. Other work
The Leeds group has provided parameters for poplar growth in the Yield-sAFe model. They
have also provided annual data on tree and crop growth (annual height, diameter at breast
height, timber volume, timber volume per hectare for trees grown with continuous cropping
and trees grown with continuous fallow; crop yield per hectare for crops in alleys and control
plots) for a paper describing the model, and have helped with the writing of the paper.
The Leeds group provided information on ranges of environmental conditions tolerable to

Populus species to Dr Riccardo de Philippi for use in modeling areas suitable for poplar
growth in WP8.
The Leeds group have provided information for the production of Deliverable 9.3 Report on
Major Silvoarable Systems of Interest in Europe.
6. Significant difficulties or delays experienced during the reporting

period
None
7. Dissemination
An interview with David Pilbeam on silvoarable agroforestry recorded at the Leeds
Experimental site was broadcast on ‘The Food Programme’ on BBC Radio 4 on 21st August
2004.
Eichhorn,, M. P., Paris, P., Herzog, F., Incoll, L. D., Liagre, F., Mantzanas, K., Mayus, M.,
Moreno, G., Papanastasis, V. P., Pilbeam, D. J., Pisanelli, A. and Dupraz, C. Silvoarable
agriculture in Europe – past, present and future prospects. Paper submitted to Agroforestry
Systems in November 2004.
De Filippi, R., Reisenauer, Y., Herzog, F., Dupraz, C., Gavaland, A., Moreno, G. and
Pilbeam, D.J. (2004). Modelling the potential distribution of agroforestry systems in Europe
using GIS. Poster presented (by R. de Filippi) at 18th International Conference for
Informatics for Environmental Protection, CERN, Geneva, 21st- 23rd October 2004.

Cranfield University Report
9 Contractor 5: Cranfield University

Partner 5 Cranfield University (CRAN, UK)
Institute Cranfield University, Silsoe, Bedfordshire, UK MK45 4DT
Group Institute of Water and Environment
Scientific team

Dr Paul J. Burgess 44-1525-863046 44-1525-863344 P.Burgess@cranfield.ac.uk
Dr Robin B. Matthews 44-1525-863008 44-1525-863344 R.B.Matthews@cranfield.ac.uk
Mr Anil Graves 44 1525 863107 44-1525-863344 A.R.Graves@cranfield.ac.uk
Mr Pascal Pasturel 44-1525-863354 44-1525-863344 P.Pasturel.s03@cranfield.ac.uk
Mr Ian Seymour 44-1525-863152 44-1525-863344 I.L.Seymour@cranfield.ac.uk
Time spent on the different work-packages

Time spent on the different work-packages months 37-42
Name WP1 WP2 WP3 WP6B WP7 WP8 Total

Technical annex 0.9 1.0 14.0 0.0 34.0 6.0 55.9
Modified (CMC1) 0.9 2.0 14.0 0.0 34.0 5.0 55.9
Year 1 0.9 0.2 0.8 0.0 10.4 2.6 14.9
Year 2 0.1 8.5 0.0 11.2 2.0 21.8
Year 3 2.7 10.5 0.0 9.6 2.4 25.2
Paul Burgess1 0.0 0.0 1.1 1.0 0.5 2.6
Anil Graves 0.0 0.0 2.2 1.8 2.0 6.0
2
Terry Thomas 0.0 0.0 0.0 0.2 0.0 0.2
Sub-total (37-42) 0.0 0.0 3.3 2.8 2.7 8.8
Total (1-42) 0.9 3.0 19.8 3.3 34.0 9.7 70.7
1
Permanent staff contribution from AC partner
2
Sub-contractor
Contribution to work-packages
WP6B. Minimal biophysical integrated model: Yield-sAFe
(Workload 3.3 person months)
The final six months were particularly challenging in terms of Cranfield University’s inputs to
the project. A key activity, not originally scheduled, has been our on-going and active
involvement in the development and calibration of the Yield-sAFe model within work-package
6B. This was necessary to provide the validated biophysical dataset that was needed for the
economic analyses within work-packages 7 and 8.
At the start of the SAFE project in August 2001, the plan was that the Hi-sAFe model would
provide such data. However in October 2003 at the CMC meeting in Orvieto, it became
apparent that the Hi-sAFe model would be unable to produce yield data for a full tree
rotation. It was therefore decided that a simplified biophysical model called Yield-sAFe
would be developed and calibrated by Wageningen University.

During 2004, Cranfield University had encoded the Yield-sAFe model into a Microsoft ©
Excel worksheet. This combination of the biophysical model with the economic model was to
allow an integrated assessment of the profitability of different forestry, agriculture and
silvoarable systems at a plot-scale. The integrated model was called Plot-sAFe. A working
version of this model with a written description of the model was made available in
September 2004 (Burgess et al. 2004).
On the basis of the calibrations undertaken by Klaas Metselaar and Karel Keesman of
Wageningen University, and reference yields established during workshops in Spain, France
and the Netherlands, Anil Graves (CRAN) and João Palma (FAL) calculated timber and crop
yields for each of the 44 land units examined in the up-scaling exercise (Table 1). These
biophysical datasets were sent to selected consortium members on 23 October 2004.
Table 111: Description of the 44 different land units and the respective assumed tree
species and crop rotation
Site Unit Rad Soil Soil depth Tree Crop rotation
(%) type (cm) species
Spain
Alcala LU1 97 Medium 140 Oak Wheat/wheat/fallow
Alcala LU2 86 Medium 50 Oak Wheat/wheat/fallow
Torrijos LU1 101 Medium 140 Oak Wheat/fallow
Torrijos LU2 100 Medium 140 Oak Wheat/wheat/fallow
Ocana LU1 100 Medium 140 Oak Wheat/wheat/fallow
Almonacid LU1 97 Medium 140 Oak Wheat/fallow
Almonacid LU2 83 Fine 140 Oak Five years of sunflower/wheat/fallow
Cardenosa LU1 93 Medium 140 Oak Wheat/wheat/wheat/fallow
Cardenosa LU2 101 Fine 140 Oak Wheat/wheat/wheat/fallow
Fontiveros LU1 99 Coarse 140 Oak Wheat/wheat/wheat/wheat/fallow
Fontiveros LU2 98 Coarse 140 Pinus Wheat/wheat/wheat/wheat/fallow
Olmedo LU1 100 Coarse 140 Pine Wheat/sunflower/fallow
Olmedo LU2 100 Medium 140 Oak Wheat/sunflower/fallow
Olmedo LU3 99 Coarse 140 Oak Wheat/sunflower/fallow
Campo LU1 99 Coarse 140 Pine Wheat/wheat/wheat/fallow
Campo LU2 99 Medium 140 Oak Wheat/wheat/wheat/wheat/wheat/fallow
Paramo LU1 100 Medium 140 Oak Wheat/wheat/wheat/sunflower/fallow
France
Champdeniers LU1 100 Fine 80 W. Wheat/wheat/sunflower/wheat/oilseed/sunflower
cherry
Champdeniers LU2 100 Medium 120 Walnut Wheat/wheat/sunflower/wheat/oilseed/sunflower
Chateauroux LU1 102 Fine 80 Walnut Wheat/wheat/oilseed/wheat/oilseed/sunflower
Chateauroux LU3 102 Medium 120 Walnut Wheat/wheat/oilseed
Chateauroux LU2 102 Fine 40 W. Wheat/wheat/oilseed/wheat/oilseed/sunflower
cherry
Chateauroux LU4 100 Fine 40 W. Wheat/wheat/oilseed/wheat/oilseed/sunflower
cherry
Fussy LU1 101 Fine 40 W. Wheat/oilseed
cherry
Fussy LU2 103 Medium 80 Poplar Wheat/wheat/oilseed
Fussy LU3 102 Fine 120 W. Wheat/oilseed
cherry
Sancerre LU1 103 Fine 40 W. Oilseed/wheat/sunflower/wheat/wheat/wheat/oilseed
cherry
Sancerre LU3 101 V fine 120 W. Oilseed/wheat/sunflower/wheat/wheat/wheat/oilseed
cherry
Sancerre LU4 100 Coarse 80 W. Oilseed/wheat/sunflower/wheat
cherry
Sancerre LU2 102 V fine 140 Poplar Oilseed/wheat/sunflower/wheat/wheat/wheat/oilseed
Champlitte LU1 103 Medium 140 W. Wheat/wheat/oilseed
cherry
Champlitte LU2 103 Md-fine 35 Walnut Wheat/wheat/wheat/wheat/wheat/grain maize
Dampierre LU1 98 Medium 140 W. Wheat/wheat/grain maize

cherry
Dampierre LU2 97 Fine 35 W. Wheat/wheat/wheat/grain maize
cherry
Dampierre LU3 95 Md-fine 60 Poplar Wheat/grain maize
Vitrey LU1 103 Medium 60 W. Wheat/wheat/oilseed
cherry
Vitrey LU2 103 Md-fine 60 Poplar Wheat/wheat/grain maize
Netherlands
Balkbrugg LU1 100 Coarse 140 Poplar Forage maize
Bentelo LU1 100 Coarse 140 Walnut Wheat/wheat/forage maize
Scherpenzeel LU1 100 Coarse 140 Poplar Forage maize
On 4 November 2004, the biophysical results were also presented by Anil Graves at the
project workshop at Zurich. At this workshop a number of concerns were raised concerning
both the parameterisation and the outputs from the model. The principal concerns were:
That the predicted timber volume of trees at a wide spacing (particularly in the 50 trees ha-1
treatment) was too high relative to those in the forest stands. Unfortunately no one was able
to present data to show whether this was actually the case.
The procedure of only changing the water-use-efficiency values to calibrate the output of the
monoculture crop and tree yields resulted in overestimates of water-use in sites with low
productivity.
The critical soil tension value beyond which crop growth was predicted to decline had been
erroneously fixed at a pF value of 2.3 (-200 cm; -20 kPa).
At or shortly after the Zurich workshop, it was proposed that these concerns could be
addressed in the following ways:
The choice of a constant light extinction coefficient of 0.8 for the tree was agreed to be too
high. Instead, particularly at low levels of tree leaf cover, the light extinction coefficient would
be reduced.
It was agreed that maximum and minimum limits should be set on the water-use-efficiency
values, and the calibration of the model to reference yields at each site could also be
achieved by modifying other factors such as the harvest index and a management factor.
A more realistic value for the critical soil water tension for crop growth would be used at a
selected pF value between 2.9 (-800 cm; - 80 kPa) and 3.3 (-2000 cm; - 200 kPa).
During November and December 2004, following the above recommendations and after making
initial assessments of the changes, Anil Graves and Paul Burgess made modifications to the
Yield-sAFe model. Between 2 and 3 December 2004, a visit to Silsoe was made by Karel
Keesman and Wopke van der Werf (Wageningen University) to discuss and agree these
changes. The changes made to the model are reported by Burgess et al (2005) in
Deliverable 6.4.
Following these changes a new complete dataset of tree and crop yields was recalculated by
Anil Graves for the 20 land units in France. These results were circulated to selected
consortium members on 11 January 2005. The datasets for the 19 land units in Spain were
recalculated and circulated by João Palma (FAL).
Christian Dupraz and Fabien Liagre, on examining the French results, again expressed some
concerns on 12 January 2005. These were:

they considered that some of the values for the land equivalent ratio were too high.
although the relative individual timber volumes at low densities had been reduced, they felt
that the predicted individual timber volumes for the widely-spaced trees that were still too
high at some sites. Again no data were available to establish whether this was correct or
incorrect.
the initial rates of timber production for wild cherry and walnut were considered to be too low.
This is despite the fact that the reference growth curves were calibrated against established
measurements for these species.
Eventually after substantial discussion, it was agreed that the Yield-sAFe predictions for
forestry, arable and the 113 trees ha-1 silvoarable system should form the basis for the
economic analysis of the landscape test sites in Spain, France and the Netherlands. The
same procedure would also be used for an analysis of the network sites. Because of a
concern that the 50 trees ha-1 datasets may overestimate tree size, it was also agreed that
these predictions would only be used provisionally.
The full dataset for the Landscape Test Sits was therefore written up by Paul Burgess, Anil
Graves (CRAN) and João Palma (FAL) and circulated as an initial draft of Deliverable 6.4 on
10 February 2005.
WP7. Economic modelling at a plot-scale

WP8. Scaling up to the farm and region

Following agreement on the Yield-sAFe modelling procedure and the biophysical dataset for the
landscape test sites on 12 January 2005, it was finally possible to undertake an economic
analysis of silvoarable agroforestry at a plot-scale.
Work-package 7 is divided into five-tasks. During the last six months of the project and
following agreement on the biophysical dataset, the focus of Cranfield University’s work has
been on tasks 7.4 and 7.5. These tasks relate to the use of the biophysical dataset to identify
the most profitable agroforestry systems at the selected network (Task 7.4) and landscape test
(Task 7.5) sites.
Task 7.4: Identification of agroforestry profitability for the network sites
Anil Graves and Paul Burgess used the Plot-sAFe model to develop the bio-physical data-
sets for the plot-scale analysis at the network sites. An initial draft of Deliverable 7.2 was
produced by Anil Graves and Paul Burgess and circulated to selected consortium members
on 24 February 2005.
Task 7.5: Identification of agroforestry profitability for the landscape test sites
Anil Graves and Paul Burgess have worked closely with João Palma to develop the bio-
physical data-sets needed for the farm-scale analysis. An initial draft of Deliverable 8.2 was
produced by Anil Graves, Paul Burgess and João Palma and circulated to selected
consortium members on 10 February 2005.

References
Burgess, P.J., Graves, A.R., Metselaar, K., Stappers, R., Keesman, K., Palma, J, Mayus, M.,
& van der Werf, W. (2004). Description of the Plot-sAFe Version 0.3. Unpublished
document. 15 September 2004. Cranfield University. 52 pp.
Burgess, P.J., Graves, A.R., Metselaar, K., Stappers, R., Keesman, K., Palma, J., Mayus, M.
& van der Werf, W. (2005). Parameterisation of the Yield-sAFe model and its use to
determine yields at the Landscape Test Sites. Unpublished report. Cranfield University.

10 Sub-contractor to CRAN : BEAM

Institute BEAM (Wales) Ltd, Gwynant, Cadnant Road, Menai Bridge, Anglesey
LL59 5BU, UK

Mr Terry Thomas 44-1248 714802 Terry.H.Thomas@virgin.net
During the last six months of the project, Terry Thomas attended and contributed to the
scientific workshop and CMC meeting in Zurich (3-6 November 2004).
WP1 WP2 WP3 WP7 WP8

Technical annex 0.0 0.0 0.0 1.5 0.0
Total first year 0.0 0.0 0.0 0.5 0.0
Total second year 0.0 0.0 0.0 0.0 0.0
Total third year 0.0 0.0 0.0 0.9 0.0
Month 37-42 0.0 0.0 0.0 0.2 0.0
Total (1-42) 0.0 0.0 0.0 1.6 0.0

CNR-IBAF Report
11 Contractor 6: CNR-Porano
Contractor 6 (CNR – Porano)

C.N.R., Istituto di Biologia Agro-ambientale e Forestale (former Istituto per l’Agroselvicoltura)
Viale G. Marconi, 2; I-05010 Porano (TR), Italy
2. Scientific team
2.1. Principal Investigators
Name Tel Fax E-mail
Pierluigi Paris +(39) 0763-374904 +(39) 0763-374980 P.Paris@ibaf.cnr.it
Francesco +(39) 0763-374903 +(39) 0763-374980 F.Cannata@ibaf.cnr.it

Cannata
Guido Bongi +(39) 075/5014538 +(39) 075/501454 G.Bongi@iro.pg.cnr.it
Andrea Pisanelli +(39) 0763-374902 +(39) 0763-374980 A.Pisanelli@ibaf.cnr.it
Giuseppe Olimpieri +(39) 0763-374908 +(39) 0763-374980 G.Olimpieri@ibaf.cnr.it
P. Paris, CNR research scientist, is the contract manager for CNR-IBAF in the SAFE Project.
During the reporting period he was involved in WP3, 4 and 2; F. Cannata, CNR research
manager, has a long experience in research management, agroforestry systems and rural
development; although he was involved with 0 man/month in the SAFE Project, he had been
offering a valuable help with his experience. G. Bongi, CNR first researcher, is a plant-
physiologist, mostly involved in studying interactions between trees and crops in silvoarables
in WP4 and 5. A. Pisanelli was contracted by CNR with SAFE funds till May 2004, as
researcher, for working on social and economical aspects of silvoarables in WP 2 and 9.
During the reporting period A. Pisanelli continued the collaboration within the SAFE Project,
but without being paid by the Project itself. G. Olimpieri is a senior technician in charge of the
technical and scientific management of the experimental fields of the CNR-IBAF at Biagio.
During the last year 2 university students were involved in the SAFE research activities. They
are both finishing the Laurea (5 years of university study) curricula in the Faculty of
Agronomy. The first student is Anna Perali, Laurea Course in Agricultural Science, at the
Università di Perugia. She is preparing for her Thesis a CD Rom in .htlm on current
silvoarable commercial plots for WP2. The second student is Alfredo Ecosse, Laurea Course
in Forestry Science at Università della Tuscia di Viterbo. He finished a Thesis on silvoarable
experimental plantations for WP3.

CNR-IBAF Report
3.Time spent on the different workpackages

0
Pierluigi Paris 1 2 1 1 5
Francesco
Cannata
Guido Bongi 1 1 2
Andrea Pisanelli 1 1 2
Giuseppe Olimpieri 2 2
4. Contribution to workpackages:
4.1. WP2: European Silvoarable Knowledge
Two main tasks were carried out during the reporting period:
Preparing a communication at a IUFRO Conference, for presenting the results of the national
survey of farmers’ view of silvoarable agroforestry systems;
Preparing a CD Rom on extant Italian silvoarable plots that were surveyed during the formers
years of the SAFE Project
4.1.1. IUFRO Presentation
After the completion of the survey of farmers’ view of silvoarable agroforestry systems and
the data analysis, the main results of the study were presented at the 7th IUFRO Extension
Working Party Symposium “Communication strategies for multiple partner involvement in
forestry extension”. The CNR-IBAF, in collaboration with IUFRO and FAO, organised the
conference in Orvieto and Rome (Italy, 27th September – 1st October 2004). The
symposium was a great opportunity for exchanging research experiences among 40
scientists coming from different countries around the world and covering the all continents.
The power point presentation of the horal communication is attached to this report as
Annex_WP2.1. .
4.1.2. CD Rom.
In collaboration with the student Anna Perali, Univeristà di Perugia, Laurea Course in
Agricultura Scienze, a CD Rom in .htlm format was prepared on the extant silvoarable
commercial plots that was surveyed within the SAFE Project during the first year of the
activity of the project. This CD Rom is part of the student’s final dissertation on the current
status of agroforestry systems in Italy. The student will defend the dissertation in April 2005.
The CD Rom is in Italian and after the mentioned date will be published on the website of the
C.N.R. I.B.A.F. .
The CD Rom is attached to this report as Annex_WP2.2. as .zip file. In order to lunch the CD
Rom, files have to be extracted by the zip format and then click on the file named
“home.htlm”.

CNR-IBAF Report
The attached version of the CD Rom is not yet the final version and some corrections and
implementations have yet to be done. One is citing, on the document itself, that the CD Rom
was realised within the SAFE Project.
4.2. WP3. Silvoarable experimental network

Aim. To collect current biophysical and economic data from the Institute existing silvoarable
experiments in central Italy, and to provide them in a unified protocol to be included in the
Database of Silvoarable Experimental Network, as required by the modelling activity.
During the reporting period, three main tasks were carried out:
Completing the measurements into two silvoarable experimental plantations of

walnuts, included into the SAFE Experimental Network, and whose characteristics
were widely described in former project reports. Both plantations were intercropped
with clover (Trifolium incarnatum L.) during the season 2003-2004 (Photo 2.
Preparing the “Tesi di Laurea” (final dissertation) of the student Alfredo Ecosse, and entitled
“Sistemi Silvoarabili: interazioni in impianti adulti tra il noce e differenti colture consociate in
due stagioni vegetative” (Silvoarable Systems: interactions between adult walnut trees and
different crops during two growing seasons). The dissertation was completed in January
2005 and will be defended on 28 February 2005, at University of Tuscia, Viterbo (Italy)
Concluding procedures for publishing a paper on the International Journal “Agroforestry

Systems”
Photo 2. Clover (Trifolium incarnatum L.) was used as intercrop in both the CNR silvoarable
experimental plantations at Biagio (Orvieto, Tr-Italy). According to our final results, clovers
are ideal intercrops in modern silvoarable systems, because clovers do not compete towards
young trees for water and soil nutrients, as well their yield is not depressed by adult trees for
light competition.
4.2.1. Experimental Measurements
The following measurements were conducted during the reporting period in the two walnuts
experimental silvoarable plantations at Biagio, (Orvieto, TR-Italy):
Growth of walnut trees;

CNR-IBAF Report
Meteo data of the year 2004;
On walnut trees: DBH, total height, bole height, crown diameter and phenology of leaf
shedding. The first three parameters were measured on December ’04, when tree had
concluded their stem growth; crown diameter was measured on September ’04, when shoots
had stopped their elongations; leaf shedding was performed from September to December
’04.
Part of the above mentioned data were analysed and they are presented in the paragraph
4.2.2.
Meteo data. Meteorological data were recorded with a local meteo-station during the
reporting period and they were acquired. Meteorological data as average monthly
temperatures and monthly rainfall are presented in the paragraph 4.2.2.
4.2.2. Laurea final dissertation.
This document is attached to the report as Annex_WP3.1. . Experimental work was carried
out in the two walnut silvoarable plantations mentioned in paragraph 4.2.1.
The dissertation document is in Italian and its extended abstract in English is reported
hereafter along with the most important data (as figures).
In Italy the rural compartment is currently undergoing dramatic changes. New land use
systems are sought in order to find solutions to main productive problems such as
agricultural surpluses, good quality timber shortage, alternative uses of marginal lands, along
with increasing environmental problems of natural resources, such as global climatic
changes, biodiversity erosion and pollutants accumulation.
Among the new proposed land uses systems, plantation forestry deserves attention for its
environmental and productive benefits. Many regulations of the European Union (Reg.
1094/88 also called “forestry set aside”; Reg. 2080/’92 for the reforestation of the agricultural
lands) have implemented plantation forestry in the member countries in the last decades.
New agroforestry systems, within the cultural models of the modern plantation forestry, may
affect significantly the sustainability of the forest plantations, increasing their multifunctional
and integration with other economical activities. Agroforestry systems have been very
common land uses systems in Italy as well in all over Europe. These systems are still
present in many areas, even if a residue of much wider ancient systems (e.g.: “coltura
promiscua” of Juglans regia in Campania, “piantate padane” in northern Italy; cereals
intercropping in olive groves in central and southern Italy).
Silvoarable systems, the intercropping of arable crops among widely spaced tree rows, could
find a special role within plantation forestry. This because silvoarable plantations could be
more profitable than forest plantations; additionally the first system can be much more
environmental friendly than the second one, decreasing soil erosion, nitrogen leaching and
forest fires.

CNR-IBAF Report
Photo 3. Young tree of walnut intercropped with lucerne (Medicago sativa L.) and mulched
with plastic film along tree line. This photo was taken in 1993, one year after tree planting, in
the silvoarable experimental plantation of Biagio 1 (Orvieto, Tr-Italy), with two walnut
genotypes (French hybrid NG23 and a common walnut half sib family). Intercropped-
mulched hybrid walnut trees in 2004 reached a stem height similar to clean cultivated-
mulched tree (see Table 112). Common walnut, intercropped and mulched, had stem growth
rates lower that clean cultivated-mulched trees (Table 113). However, the used genotype of
common walnut showed not a perfect adaptability to site conditions of the experimental site.
Much better results were obtained for intercropped common walnut trees of the variety
Feltrina (see Table 114)(from Northern Italy) in the other adjacent silvoarable experimental
plantation (Biagio 2) The choice of the right genotype for common walnut is a crucial factor
for a successful establishment and development of the plantation.
Since the early ‘90s in the experimental fields of the Istituto di Biologia Agroambientale e
Forestale (IBAF) of Porano, Italian National Research Council, research activity has been
carrying on silvoarable systems with walnuts (Juglans spp.) trees intercropped, since
planting time, with typical crops of peninsular parts of Italy, such as wheat and fodder crops.
Walnut trees were either mulched (with plastic film) or unmulched during the first years sinc
planting time () This was in order to make it easier weed control close to the tree base, as
well to reduce possible intercrop competitions towards young trees.
The present “Thesis of Laurea” was developed in collaboration with the IBAF-CNR and within
the European Project “S.A.F.E.” (Silvoarable Agroforestry for Europe) (2001-05), which
includes a consortium of European Institutions and with the coordination of the INRA,
Montpellier, France (www.montpellier.INRA.fr/safe).
The main general objective of the thesis investigation is studying the interrelations between
adult walnut trees and two intercrops, wheat (Triticum aestivum L.) and clover (Trifolium
incarnatum), during two growing season (2003 and 2004) in two experimental walnut
plantations established in 1992 and 1994, respectively, in the hilly area of Monti Vulsini
(Orvieto, Italy), an area of volcanic origin and with a meso-Mediterranean climate (Photo 4;
Table 112).

CNR-IBAF Report
Photo 4. Winter view (January ‘05) of one of the two experimental silvoarable plantations of
walnuts at Biagio (Orvieto, Tr-Italy). The climate of the area is sub-humid Mediterranean, with
average annual precipitation of 850 mm, and two months of summer drought (July and
August). These conditions are optimum for walnut in Italy for timber production.
Biagio 2003: Total rainfall 653.8 mm; Biagio 2004: Total rainfall 1263 mm;
Average temperature 11.06 °C Average temperature 11.4 °C
160 80
Monthly rainfall 2003 (mm) 250
Monthly temperature (°C)
Monthly rainfall 2004 (mm) 120
Monthly temperature (°C)

140 70
Monthly rainfall (mm)
Monthly rainfall (mm)
Average Temperature 2003 (°C) Average temperature 2004 (°C)

120 60 200 100
100 50 80
150
80 40 60
100
60 30
40
40 20
50
20
20 10
0 0
0 0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Table 112. Thermo-pluviometric diagram at Biagio (Orvieto-TR, Italy) during the last two
years.
One of the limiting factors for intercrops in adult silvoarable plantations is light that may be
highly in shortage due to the shade of the large tree crowns. Therefore, it is important to
select shade tolerant intercrops and also to be able to predict crop yield depression
according to tree age and spacing. With this aim, measurements were carried on wheat and
clover yields and growth, and these parameters were related to the main tree growth
parameters (DBH, H, crown diameter, leaf phenology, basimetric area-G), as well with the
solar radiation (PAR) intercepted by tree foliage (with hemispherical photos).
In general, the results are highly promising in favor of the silvoarable plantations of walnut,
with stem growth of intercropped trees similar to that of forest plantations, especially taking
particular attention to the selection of walnut genotype that must be adapted to the local
pedo-climatic conditions (Table 113; Table 114).

CNR-IBAF Report
DBH (cm ) n.s

20
15 Biagio 1-Hybrid Walnut
14 10 14
5
0 n.s
12 12
CC-mulched A F-mulched
DM-crop yield (t ha-1)

10 10
Stem H. (m)
8 8
6 6
4 4
2 2
0 0
1992 1994 1996 1998 2000 2002 2004
Lucerne W.Wheat Clover CC-Mulched AF-Mulched
DBH (cm )
*
20 Biagio 1-Com m on W alnut

14
15 ** 14
10
5
12 0 12
D M -crop yield (t ha- 1)

C C-M ulched A F -m ulc hed
10 10
S tem H. (m )
8 ** 8
6 6
4 4
2 2
0 0
199 2 1 994 1 996 1 998 20 00 200 2 200 4
Luc ern e W.Wh eat Clo v er CC-Mulc h ed A F- Mulc hed
Table 113. Biagio 1, Orvieto (Tr). Total stem height (Stem H) and diameter (DBH) of
hybrid and common walnut for CC-mulched and AF-mulched since planting time.
Vertical bars represent ± s.e.m.. ** = highly significant difference for p ≤ 0.01 (t.Test).
n.s = not significant difference.
9
Biagio 2-Common walnut
8
A A
7
B
6
Stem H (m)
B
5
4
3
2
1
0
1994 1995 1996 1997 1998 1999 2001 2002 2003 2004
Years
Mulched-clover Grassing-dow n
Clover Clean-cultivation
Table 114. Biagio 2, Orvieto (Tr). Total stem height (Stem H) of the common walnut
from 1992 to 2004. The four different lines represent the different thesis. Vertical bars
represent ± s.e.m. Different letters mean significantly differences for p ≤ 0.01 (L.S.D.
Test).
Comparing the collected data with former ones, current results show the importance of the
management of the tree-crop interface, and also the changes of tree-crop interactions

CNR-IBAF Report
according to the intercrop. Clover resulted less competitive than wheat towards walnut young
trees and also much less affected by walnut shading in adult plantations. Therefore clover
can be used as intercrop with large-adult trees whose shade do not hinder clover yield,
contrary to what observed for wheat (Table 115).
B i a g i o 1-200 3 (W he a t)
6
Ne t crop y ield (t ha )
-1
5
0
Common Ne w A F Hy brid New A F Sole c r op
mulc hed mulc he d
Bia g io 1-200 4 (Clov e r)

7
Ne t cro p y ie ld (t ha )
-1
6
5
4
3
2
1
0
Common New A F Hybr id New A F So le c ro p
mu lched mu lched
Table 115. Biagio 1, Orvieto (Tr). Net crop yield of wheat (2003) and clover (2004)
intercropped with walnut. Vertical bars represent ± s.e.m
This because clover growth occurs mostly before walnut trees completed the full emission of
leaves in spring (Table 116). On the contrary, wheat is a late crop, whose harvesting
occurred a month later than walnut full leaves emission (Table 116). Furthermore, clover is
also useful for soil erosion protection, and doesn’t depress tree growth in comparison with
B ia gi o 1 : W a l nu t & w he a t, 2 003
Threshing
90 3, 5
of Wheat
80
3, 0
70
2, 5
60
H crop (cm )
2, 0
T re e Ph. I.
50
40 1, 5
30
1, 0
20
10 0, 5
0 0, 0
25 /3 1/ 4 9/ 4 15 /4 23 /4 29 /4 6/ 5 13 /5 20 /5 2 8/ 5 3/ 6 1 7/ 6 2 4/ 6 8/ 7
D a te (dd/m onth)
H .inter c r op H Sole c r op BB Common W alnut B B Hy br id W alnut

CNR-IBAF Report
B ia g i o 1 : W a l nu t & cl o ve r, 2 004
10 0 3, 5
90
3, 0
80
70 2, 5
H crop (cm )
T ree Ph . I.
60 2, 0
Mowing
50
of 1, 5
40
Clover
30 1, 0
20
0, 5
10
0 0, 0
24 / 3 1/ 4 8 /4 8/ 4 21/ 4 28 / 4 6/ 5 26/ 5 3/ 6 15/ 6 22 /6 29/ 6 8 /7
Da te (dd/m onth)
H.inter c r op H Sole c rop BB Hy brid W alnut BB Co mmon W alnut
Table 116. Biagio 1, Orvieto (Tr). Relationship between hybrid and common walnut
buds break and herbaceous crop height (intercrop and sole crop). Vertical bars
represent ± s.e.m.
grassing down with endemic grass sward (Table 114).
These observations were enforced by those ones with hemispherical photos, which showed
consistent relations between the PAR, available underneath walnut crowns, and intercrop
yield depressions (in comparison to the sole crop) according to the two different crops.
Intercropped wheat decreased significantly its yield with increasing tree size (see Table 118
in the paragraph 4.4). Intercropped clover yield was unaffected by tree size (see Table 119,
in the paragraph 4.4). Furthermore, the plantation basimetric area (G) may be a very useful
parameter for a first estimation of light competition of walnut trees towards intercrops,
according to tree growth and spacing. There was a significant negative linear regression
between G and wheat yield depression underneath tree shade. The higher is G and the
lower is the yield of the intercropped wheat. G increases with increasing tree stem
dimensions (because there is a linear positive relation between stem DBH and crown
diameter (Table 117), but G decreases with increasing tree spacing. According to our
simulation, optimum tree spacing in walnut silvoarable systems could be 10 x 14m. With this
spacing and with stem growth rates of 1 cm/year, intercropped wheat yield depression never
drops to 50% of reference yield before walnut harvesting time.

CNR-IBAF Report
B ia g io 1
10
9
y = 0,278 9x + 1 ,6 655
8
D crow n (m ) 7 R 2 = 0 ,7 173 **
6
5
4 y = 0,3437 x + 0,52 99
3
2 R2 = 0 ,8 431 **
1
0
0 5 10 15 20 25 30
DB H (cm )
Hybrid Waln ut C om m on Wal nut
Table 117. Biagio 1, Orvieto (Tr). Relationship between crown average diameter (D
crown) and stem diameter at breast height (DBH) of walnut. ** = highly significant for p
≤ 0.01.
As general conclusion, considering the data of the study sites, indications are that walnuts
silvoarable systems could have individual tree growth rates similar to that of forest
plantations, and yield of intercrops may be unaffected by tree shading according to crop type
(warm versus cool season crops) and tree spacing (close versus open plantations). These
indications are valid for sites with pedo-climatic conditions being optimum for walnut, and
using walnut genotypes well adapted to the site conditions.
4.2.3. Paper in press
During the reporting period CNR-Italy Partner concluded all the final steps for publishing the
following paper:
Paris P., Pisanelli A., Todaro L., Olimpieri G., Cannata F., 2004. Growth and water relations
of walnut trees (Juglans regia L.) on a mesic site in central Italy: effects of understorey herbs
and polyethylene mulching. Agroforestry Systems (in press)
This paper was already mentioned in the former SAFE report as accepted by the journal’s
editor after minor revisions. During this last semester we had performed the requested
revisions and reviewed the print proof, that is attached to this report as Annex_WP3.2. .
4.3. WP4 Above ground interactions

During the reporting period we carried out activities in the experimental fields 1 and 2 of
Biagio (see WP3 report), concerning in analysing data already recorded or measuring the
leaf shedding of walnut trees in the period September-December 2004.
Walnut leaf shedding data of the year 2004 were collected according to the common SAFE
protocol for tree leaf phenology. These data have not yet analysed.
Formerly collected data were analysed and put in relation with the growth of intercrops (as
plant height) during the years 2003 and 2004. These data are shown in the paragraph 4.2.2
as Table 116.

CNR-IBAF Report
More in general extent, in conclusions to our research activity concerning the above-ground
interactions between trees and intercrops for light competition, we observed that with the
narrow tree spacing (6 x 7 m) of one experimental walnut plantation (Biagio 1), the wheat
yield was negatively affected by increasing tree size (Table 118); on the contrary, clover yield
was not affected by increasing tree size (Table 119); this because clover is a cool season
crops whose growth occurred much before walnut full leaves emission (Table 116).
B ia g io 1-G ro w in g se a so n 2003: W h e a t
120
% of crop reference yield

100
80
WHEAT
WHEAT 60
y = -7,0056x + 94,982
40
R 2 = 0,392 *
Decreasing wheat yield with 20
y = -8,3879x + 107, 88
R 2 = 0,6492 **
increasing tree size 0
0 1 2 3 4 5 6
G (m 2 h a -1 )
Hy brid w a lnut Common w aln ut
Table 118. Biagio 1, Orvieto (Tr). Relationship between walnut stem basimetrical area
(G) and decreasing production of intercropped wheat in comparison with sole wheat. *
= significant for p ≤ 0.05, ** = highly significant for p ≤ 0.01.
Biagio 1-Grow ing season 2004: Clover
120
% o f crop reference yie ld
100
80
60
40
CLO V ER y = -2,1044x + 88,661 y = 1,8445x + 77,88
20 2 2
R = 0,0517 ns R = 0,0527 ns
0
0 2 4 6 8
2 -1
Constant clover yield with increasing trees G (m ha )
size C om m o n Wa ln ut H ybrid w alnut
Table 119. Biagio 1, Orvieto (Tr). Relationship between walnut stem basal area (G) and
intercropped clover yield in comparison with sole clover. G is function of tree stem
diameter (DBH) and tree spacing; being tree spacing constant, as in most of our
experimental plots, G varies with DBH or tree size. n.s = not significant.
Similar results were obtained for the relationship between intercropped clover yield and the
stand basal area (G), or tree size, in the experimental plantation Biagio 2 during the year
2004 (data not shown in this report).
It must be stressed out that in Biagio 1 silvoarable plantation we used a tree spacing of 6 x
7m that is definitively too low for adult silvoarable plantations. This tree spacing was used
just for experimental reason, in order to have treatments replications within a reasonable
experimental area.

CNR-IBAF Report
4.4. WP9 EU guidelines

This report summarizes the current status of silvoarables systems in Italy and how this was
affected by recent regulations of the Common Agricultural Policies (CAP) of the European
Union.
According to the research activity carried out during the past months within the WP2, the
current status of silvoarable system in Italy is reported in Table 120. In most cases the area
covered by each of the listed systems is not reported in official statistics and the values in the
table are the results of personal estimation of the authors of this report. Specific research
was activated at ISTAT, Istituto Nazionale di Statistica, the main Italian reference for
statistics in agriculture and forestry. Unfortunately this investigation was not successful,
because silvoarable systems are completely denied.
Silvoarable system Area (ha) Source
Intercropped olive groves 200,000 p.e.
Arable lands with scattered oaks (also with Cork 180,000 p.e.
oak in Sardinia Island)
Lapietra et
al.1991.
Hybrid Poplars plantations 12,500
Walnut groves 5,000 p.e.
Table 120. Main extant silvoarable systems in Italy. P.e.: personal estimation.
In Table 121, pictures of those systems are reported along with their geographic distribution.
Some system are very local and linked to very peculiar site and socio-economic conditions,
other systems are more common all over the national territory (and over). Local systems are
the intercropped cork oak system that is mostly present in Sardinia Island, where cork oak
has its climatic optimum; also the walnut system is local and mostly present in Campania
Region, where site conditions are extremely good (volcanic flat soil, warm-humid climate).
Common systems are the oaks systems, the olive system (just for the peninsular part of Italy)
and the hybrid walnut plantation systems (in the Po Valley). Scattered walnut in cropland can
be found all over Italy.
Indifferently whether local or common systems, they all have had a strong decline during the
second half of the XX century, because of the modernisation of agriculture and the linked
socio-economic changes. An example of those declines is presented in Table 122 for walnut
silvoarable system in Italy. Till a recent past the species was much common in the Italian
countryside and the prevalent form of management was the intercropping of widely spaced
trees with vegetables, tobacco and fruit trees, in very fertile areas, and scattered trees in
cropland of wheat or fodder crops in less fertile areas.

CNR-IBAF Report
Those silvoarable walnut systems rapidly declined as for many other systems incorporating
trees into farmland. Most of these trees were valuable hardwoods (Fraxinus, Acer, Prunus)
that have completely disappeared from croplands. Valuable timber is currently in strong
shortage on the Italian market, and industries import large volume of tropical timbers.
Among the reasons of the silvoarable systems decline there are also the European CAP and
the national agricultural and forestry policies. These highly supported crops production with
generous subsidies, and farmers, in order to maximise crop yield and harvesting of grants,
eradicated most of the scattered trees into agricultural fields. In many cases these trees are
protected by several laws (with environmental and panoramic issues), but this was not
sufficient to stop the decline of agroforestry systems, because no new young trees are
planted or grow naturally.
Table 121. Pictures showing the main silvoarable systems in Italy and their
distribution within the nation territory (blue arrows). Local systems are in yellow,
whereas more general systems are in red. The red straight lines separates the
continental part of Italy, with a sub-continental climate, from the peninsular par, with
various degree of Mediterranean climate.
Farmers avoid any regeneration of trees into farmland because the area covered by each
tree is subtracted by the total cropland area, decreasing the crops subsidies (see Table 123).
Furthermore, no subsides were never activated by the CAP for scattered trees into cropland.

CNR-IBAF Report
silvoarable
140.000
pure
Area (ha)/ Fruit (x100 Kg)

120.000 Total Area
100.000 Nut production
80.000
60.000
40.000
20.000
0
1962 1967 1972 1977 1982 1990
Table 122. Decline of multipurpose (fruit and timber) cultivation of common walnut in
Italy during the second half of the XX century. Silvoarable groves were the major
management system for this tree species.
Table 123. Different systems for calculating cropland surface due to the area occupied
by trees in silvoarable systems. Calculation systems differ according to administrative
Region, tree species and variety. Cropland reduction is calculated dividing the current
number of trees over the value indicated in the last column (n°/ha). Example with 40
olive trees ha-1, cropland is reduced of 1700 m2 per ha for low vigorous variety in
Lazio.
Limit of tree
Tree species Admin. Region Tree variety density per ha
(n°/ha)
Olive tree Lazio Low vigorous 230 ha-1
High vigorous 150 ha-1
Common walnut Campania
Cork oak Sardegna
The only silvoarable system with double summed tree and crop subsidies is the intercropped
olive system (see Table 124). In this case, in Italy, till the last CAP Reform, crops subsidies
were managed on a surface basis (per ha), and olive subsidy according to total olive
production. The area covered by olive tree is subtracted to the crop area, according to rules
applied by each administrative Region and olive tree variety.

CNR-IBAF Report
Table 124 Main production of extant silvoarable systems in Italy, European subsidies
and future trend of systems according to the current Common Agricultural Policy
(CAP) of the European Union. Legend: Products in bold have been subsidized by CAP
* for cork oak.
Tree component Current

status
Silvoarable
Crops Products Subsidies for Future trends
Systems Laws for for crop double
protection area grants
Fuelwood Timber Fruit reduction
Olive X X Yes No Yes Stable

system
Oaks x Cork* X Yes No No Slow decline

systems
Poplar x X No No No Rapid declining
system
Walnut x X X No No No Rapid declining

systems
Since this January, the new CAP reform is in force in Italy. This reform is called Single Farm
Payment (SFP), and European agricultural subsides are not anymore linked to the crop type
and its yield; new subsidies will be devoted to the farmer on an hectare basis and according
to an average value calculated using the past subsides. In this case, scattered trees into the
cropland will be anymore subtracting crop area for European subsides. Although that, most
of the farmers will continue to avoid new trees growing within their cropland. For these
farmers trees depress most of the crop yield unless trees can produce marketable products
like olive trees. According to this, the future trend is stable just for the olive silvoarable
systems, and declining for the remaining systems (Table 124). For the oaks systems the only
way to invert the negative trend would be subsidising trees planting into cropland. This
support should be based mostly on the environmental benefits of having scattered oaks trees
into farmland. These environmental benefits (soil erosion protection, nutrient leaching
reduction, biodiversity improvement) have not been fully investigated by the SAFE Project
(by FAL Unit-Zurich, Switzerland) and deserve further investigation.
For the silvoarable systems with timber trees (hybrid poplars and walnuts) the trends is a
rapid decline, according to the current situations. Two strategies deserve attention to reverse
the negative trend. One should be producing convincing examples demonstrating their higher
profitability in comparison to pure plantations and pure crop. The SAFE Project constructed
simulation models to compare profitability of silvoarable systems in comparison to pure forest
or crop systems. Unfortunately simulations results of SAFE models are not yet available. In
any case, according to our experimental results (see WP3 in this report) the “indications are
that walnuts silvoarable systems could have individual tree growth rates similar to that of
forest plantations, and yield of intercrops may be unaffected by tree shading according to
crop type (warm versus cool season crops) and tree spacing (close versus open
plantations)”. These conclusions are in favour for a good profitability of silvoarable systems
with timber trees.

CNR-IBAF Report
A second possibility for reversing the decline of silvoarable systems would be supporting tree
planting, at low density, into croplands without stopping subsides for crops. This second
options should be considered also in consideration of the past results of European
regulations for tree planting in alternative to surplus crops (EU Reg. 2080/’92). Unfortunately,
in Italy results of the plantations established with this regulation were not always very
successful. This was because farmers devoted to tree planting just the worse land, where
tree establishment is very difficult. Supporting tree planting according to agroforestry
schemes (intercropping with crop) would certainly stimulating farmers to plant trees into good
cropland, where trees like poplars, walnuts, cherry and other valuable hardwoods can grow
well and produce good quality wood that is strongly in shortage on the European market.

University of Extramadura Report
12 Contractor 7: University of Extramadura

Contractor 7 : School of Forestry, University of Extremadura (UEX, Spain)

Centro Universitario, University of Extremadura (UEX)
10600, Plasencia, Spain

Name E-mail
Gerardo Moreno gmoreno@unex.es
Fernando Pulido nando@unex.es
Pilar Rubio pilar@unex.es
Abelardo García abgarcia@unex.es
Julio Hernández juliohb@unex.es
Juan C. Giménez jcfernan@unex.es
Mercedes Bertomeu bertomeu@unex.es
Elena García Cubera ecubera@unex.es
María Jesús Montero cmontero@unex.es
Parejo
Time spent on the different workpackages during the third year
Gerardo Moreno 0,5 1 2 1 0,5 0,2 0,5 1 6,7
Fernando Pulido 1 1 2
Juan Carlos Giménez 0,3 0,3
Mercedes Bertomeu 0,3 2 0,3 2,6
Julio Hernández 0,0
María Jesús Montero 0,2 1 1,5 1,5 0,2 4,4
Elena Cubera 1 3,5 4,5
Maite Bellido 2 2
Subcontrators
Eduardo de Miguel 0,3 0,3
SubTotal (VIII/03 – VII/04) 0 3,0 4 4,5 6 0,8 2,4 1,1 1 0 22,8
Total (VIII/01 – VII/04) 0 7,3 14 10 14,3 0,8 3,5 5,5 2 0 57,4
Foreseen person-months
in Technical Annex 0 5 12 3 9 0 3 2 2 0 36
Difference (next 6 months) 0 +2,3 +2 +7 +5,3 +0,8 +0,5 +3,5 0 0 ----
SUMMARY
In this third year we have focused our effort on the divulgation and publication of the
preliminary results of our contribution in the SAFE project. We have elaborated 6 scientific
publications to be published in international journal (Annex 1) and we have participated in an
International Congress on Sustainable Forest Management, with two communications
(Annex 2). Moreover, we have participated in national/regional workshops for discussion of
the current situation of Agroforestry in Spain and the perspectives with the new CAP. We
have also prepared a four-page pamphlet to support the divulgation of Agroforestry practise

in Spain (Annex 3). Presently we are working in the organization of the national end-users
meeting, which will be carried out next 11th march in Madrid.
In addition, we have collaborated with responsible of the WP7 and WP8 to run and to
analyse results of the Yield-sAFe, for Spanish experimental farms and LTSs (Landscape
Test Sites). These results will be delivered in the “Report on the plot-economics of European
silvoarable systems” (Deliverable 7.2) and in the “Report on Economic feasibility of
silvoarable agroforestry in target regions” (Deliverable D8.2). We have also prepared a
document about the taxation of agroforestry systems in Spain: “Forestry, Agroforestry and
Taxation issues in Spain” (Annex 4).
Finally, UEX team has participated in the Final Scientific Meeting of the SAFE Project and
the CMC meeting both Zurich, October 2004. Elena Cubera made two working visits to
INRA-Montpellier in February and July 200413 (a brief description of her activity is given in
annex 5).
PERSON-MONTHS
Time spent on the different work packages during the third year
Gerardo Moreno 0,2 0,5 1 1 0,1 0,1 0,5 3,4
Fernando Pulido 0,5 0,5
Mercedes Bertomeu 0,2 0,2
Elena Cubera 2 2,0
Juan Carlos Moreno 1 1,5 2,5
Eduardo de Miguel
0,5 0,5
(Subcontractors)
SubTotal (VIII/04 – I/05) 0 0,2 0,5 2 5 0 0,3 0,1 1 0 9,1
Total (VIII/01 – I/05) 0 7,5 14,5 12 19,3 0,8 3,8 5,6 3 0 66,5
Foreseen person-months
0 5 12 3 9 0 3 2 2 0 36,0
in Technical Annex
+ + +
Difference 0 + 2,5 + 2,5 + 0,8 + 0,8 + 3,6 0 +30,5
9 10,3 1
WP2. European silvoarable knowledge

We have just collaborated on the preparation of the paper entitle “Silvoarable systems in
Europe – past, present and future prospects”, submitted for its publication in the scientific
journal “Agroforestry systems”. This paper is included in the report of Partner 4 (University of
Leeds); it is not included in our report to avoid redundancy.
On the other hand, in this period has been published the paper “El árbol en el medio
agrícola” in Foresta (nº 27: 70- 76), a national journal specialized in forestry (This paper was
already included, as manuscript, in the third year report).
WP3. Silvoarable experiment work

In this period UEX have prepared a synthesis of the most prominent results of the four
experimental farms. From here, we are preparing a paper to be published in a special issue
13
These working visits are mentioned here because it was not included in the previous
report, the third year report. On the other hand, the participation of Gerardo Moreno in the
First World Agroforestry Congress was reported in the third year report, but the costs are
included in this last report.

of the Agroforestry systems journal, which will include a selection of papers presented in the
International Congress “Silvopastoralism and Sustainable Management”, carried out in Lugo
– Spain, April 2004 (see the third year report). A first draft of this paper is here included
(Annex 1.2).
A. Publications
Gerardo MORENO, José Jesús OBRADOR, Eustolia GARCIA, Elena CUBERA, María Jesús
MONTERO, Fernando PULIDO and Christian DUPRAZ. 2005. Competitive and Facilitative
interactions in dehesas of C-W Spain. First draft of a paper to be published in a special issue
of the Agroforestry systems journal. (Annex 1.2).
WP4. Above-ground interactions
UEX has continued with some of the experiments initiated in previous years and commented
in previous reports. In short, we can resume these activities in the following list:
1. To measure the physiological status of the tree (sap flow velocity, water potential and
photosynthesis) in function of the land use and tree density (study still in course).
2. To determine the fruit (acorn) production (Finished).
3. Emergence and survival of seedlings (Finished).
On the other hand, we have written three papers (Annex 1.3 to 1.5). Here, we just list the title
of this works.
Publications
Gerardo MORENO, José Jesús OBRADOR, Abelardo GARCÍA. 2005. How trees determine
soil nutrient distribution in intercropped dehesas: consequences for crop production.
Paper submitted to Agroforestry Systems. (Annex 1.3).
María Jesús MONTERO and Gerardo MORENO. 2005. Light distribution in scattered-trees
open woodland in Western Spain. Paper submitted to Agroforestry Systems. (Annex 1.4).
Fernando PULIDO et al. 2005. Acorn production and regeneration of holm oak dehesas as
influenced by under-storey management. Working document for three future papers to be
submitted to the journals Ecology, Journal of Applied Ecology and Agroforestry systems.
(Annex 1.5).
WP5. Below-ground interactions
UEX has continued with some of the experiments initiated in previous years and commented
in previous reports. In short, we can resume these activities in the following list:
To go on with soil water dynamic measurements (study still in course).
We have followed up with the experiments to study the root dynamic of poplar and holm-oak,
in order to parametise the cellular root model developed Rachmat Mulia and Christian
Dupraz in the context of the SAFE project (study still in course).
On the other hand, this year we have presented two communications to International
Congress (Annex 2.1 and 2.2) and we have written a paper (Annex 1.6).
Communications to Congress

Gerardo MORENO and José Jesús OBRADOR. 2004. Consequences of dehesa land use on
nutritional status of vegetation in Central-Western Spain. International Symposium “Forest
soils under global and local change: from research to practice”, which was carried out in
Bordeaux (France) in September 2004. (Annex 2.1).
Elena G and Obrador JJ. 2004. Consequences of dehesa land use on nutritional status of
vegetation in Central-Western Spain. International Symposium “Forest soils under global
and local change: from research to practice”, which was carried out in Bordeaux (France)
in September 2004. (Annex 2.2).
Publication
Gerardo MORENO and José Jesús OBRADOR. 2004. Soil nutrient content and nutritional
status of holm-oaks in dehesas of C-W Spain. Paper submitted to Annals of Forest
Science (Annex 1.6).
WP8. Scaling-up to the farm and the region
In this period UEX has revised some economical and physical data collected to run Yield-
sAFe, Plot-sAFe and Farm-sAFe models for different situations in Spain (9 LTSs selected in
the WP 8).
This task has been made in close collaboration with Anil Graves and Paul Burgess (Cranfield
University), and Joao Palma (FAL), who is implementing the models. Therefore, results of
economic simulations for Spanish scenarios are presented in the “Report on the plot-
economics of European silvoarable systems” (Deliverable 7.2) and in the “Report on
Economic feasibility of silvoarable agroforestry in target regions” (Deliverable D8.2), both
reports prepared by CRAN and FAL partners. Therefore they are not included in our report to
avoid redundancy.
WP9. European guidelines for policy
In these last six months, UEX has prepared a report about the taxation of agroforestry
systems in Spain: “Forestry, Agroforestry and Taxation issues in Spain” (Annex 4). This
report is an addendum to the document Eligibility of Silvoarable Systems for Government
Financial Support, which was presented in the third year report, in the annex 9.2.
In addition, we have participated actively in the promotion of the discussion of the present
situation of Agroforestry in Spain and the implications of the new CAP, with special attention
to the consequences of the single payment scheme and the proposal Rural Development
Regulation (2007-2013). UEX was presenting SAFE results and promoting the agroforestry
practice in the following meeting:
Workshop on Silvoagricultura, cultivo e inversión, Soria, 28-29 October 2004. Organized by

CESEFOR-Junta de Castilla y León. Participation of Gerardo Moreno as invited speaker:
Funcionalidad, manejo y estado actual de la Silvoagricultura.Congreso PAC, Modelo
Agrario Europeo y Sociedad. Cáceres 10-12 November. Organized by Junta de
Extremadura y Ministerio de Agricultura y Pesca. Participation of Gerardo Moreno.
Technical meeting Especies Forestales Productoras de Madera de Calidad, Aguilar de

Campoo (Castilla-León), 11-12 November 2004. Organized by Federación ADEMPA.
Participation of Gerardo Moreno as invited speaker: “Silvoagricultura, otra forma de hacer
agricultura, otra forma de hacer agricultura.Workshop Las medidas de Desarrollo Rural de
la PAC y la conservación de la naturaleza en Extremadura, Jarandilla de la Vera, 14-15
de December 2004. Organized by the Instituto de Desarrollo Rural Sostenible IDRiSi.
Participation of Gerardo Moreno as invited speaker

National meeting for the forestation program (National and Regional Administrations in
charge of the aforestation programs). Ciudad Real, 15th February. Participation of
Gerardo Moreno as invited speaker: Perspectivas para la Silvoagricultura en Europa.
Finally, we are preparing the end-users meeting, which will be next 11th march in Madrid, and
around 100 are attended. In this meeting results of the SAFE project will be presented and
the policy aspects affecting agroforestry in Spain and Europe will be discussed.
Sub-contracted work to FGN

The major contribution of FGN has been to provide a silvoarable plot to carry out the most
part of the experiments of the project.
In this moment they are collaborating in the organization of the ‘end-users meeting’, in
Madrid, where they have its main offices.
Exploitation and dissemination activities

In these last six months we have concentrated important effort on the dissemination of the
result of the SAFE project.
We have participated in five (national/regional) meetings (quoted in WP 9), with the

participation of end-users and the agriculture and forestry administration, presenting and
discussing the opportunities for a new Agroforestry in Spain, and the following challenges
with the new CAP. In addition, we have edited a four-page pamphlet describing the main
characteristics of the agroforestry systems arising from the results of the SAFE project
(Annex 3). This pamphlet has been distributed all around Spain through farmer and forester
associations and national and regional administrations.
Apart of the dissemination of the results for end-users and stakeholders, we have made a
scientific dissemination of our results in an international congress in Bordeaux (France), in
September 2004 with two communications (Annex 2), and we have prepared 6 papers, which
are being sent to different scientific journal (Annex 1). In this period has been also published
three paper that were already included in the third year report as manuscript, thus they are
not included now to avoid redundancy. These papers are:
Cubera E., Montero M.J. and Moreno G. 2004. Effect of land use on soil water dynamics in
dehesas of Central-Western Spain. In: Schnabel S. and Ferreira A. (eds) Advances in
GeoEcology 37: Sustainability of Agrosilvopastoral systems –Dehesas, Montados-.
Catena Verlag, Reiskirchen, Germany, pp 109-123.
Montero MJ, Obrador JJ, Cubera E and Moreno G (2004) The role of dehesa land use on
tree water status in Central-Western Spain. In Advances in GeoEcology 37: Sustainability
of Agrosilvopastoral systems –Dehesas, Montados-. Eds. S Schnabel and A Ferreira. pp.
125-136. Catena Verlag, Reiskirchen.
Obrador-Olán, J.J., García-López, E., & Moreno, G. (2004). Consequences of dehesa land
use on nutritional status of vegetation in Central-Western Spain. In Advances in
GeoEcology 37: Sustainability of Agrosilvopastoral systems –Dehesas, Montados-. Eds. S
Schnabel and A Ferreira. pp. 327-340. Catena Verlag, Reiskirchen.
List of annexes
ANNEX 1. Publications produced by UEX group and in collaboration with other SAFE
members in the period August 2004-January 2005.

ANNEX 2. Relation of communications presented to international and national congress by

UEX group in the period August 2004-January 2005.
ANNEX 3. Pamphlet on Agroforestry.
ANNEX 4. Report “Forestry, Agroforestry and Taxation issues in Spain”.
ANNEX 5. Report of the working visit of Elena Cubera to INRA-Montpellier.
ANNEX 1 Publications (August 2004-January 2005)
ANNEX 1.1. Gerardo Moreno. 2004. “El árbol en el medio agrícola” in Foresta, 27: 70- 76.
ANNEX 1.2. Gerardo MORENO, José Jesús OBRADOR, Eustolia GARCIA, Elena CUBERA,
María Jesús MONTERO, Fernando PULIDO and Christian DUPRAZ. 2005. Competitive
and Facilitative interactions in dehesas of C-W Spain. First draft of a paper to be
published in a special issue of the Agroforestry systems journal. (Annex 1.2).
ANNEX 1.3. Gerardo MORENO, José Jesús OBRADOR, Abelardo GARCÍA. 2005. How
trees determine soil nutrient distribution in intercropped dehesas: consequences for crop
production. Paper submitted to Agroforestry Systems. (Annex 1.3)
ANNEX 1.4. María Jesús MONTERO and Gerardo MORENO. 2005. Light distribution in
scattered-trees open woodland in Western Spain. Paper submitted to Agroforestry
Systems. (Annex 1.4)
ANNEX 1.5. Fernando PULIDO et al. 2005. Acorn production and regeneration of holm oak
dehesas as influenced by under-storey management Working document for three future
papers to be submitted to the journals Ecology, Journal of Applied Ecology and
Agroforestry systems.
ANNEX 1.6. Gerardo MORENO and José Jesús OBRADOR. 2004. Soil nutrient content and
nutritional status of holm-oaks in dehesas of C-W Spain. Paper submitted to Annals of
Forest Science
A last paper it is included in the report of Partner 4 (University of Leeds). M.P. Eichhorn, P.
Paris, F. Herzog, L.D. Incoll, F. Liagre, K. Mantzanas, M. Mayus, G. Moreno, V.P.
Papanastasis, D.J. Pilbeam, A. Pisanelli and C. Dupraz 2004 “Silvoarable systems in
Europe – past, present and future prospects”, submitted for its publication in the scientific
journal “Agroforestry systems”.
ANNEX 2 Communications (August 2004-January 2005).
Gerardo MORENO and José Jesús OBRADOR. 2004. Consequences of dehesa land use on
nutritional status of vegetation in Central-Western Spain. International Symposium “Forest
soils under global and local change: from research to practice”, which was carried out in
Bordeaux (France) in September 2004. (Annex 2.1).
Elena G and Obrador JJ. 2004. Consequences of dehesa land use on nutritional status of
vegetation in Central-Western Spain. International Symposium “Forest soils under global
and local change: from research to practice”, which was carried out in Bordeaux (France)
in September 2004. (Annex 2.2).

FAL (Swiss federal research station fur agroecology and agriculture) Report
13 Contractor 8: FAL
Contractor 8 : FAL (Switzerland)

: Swiss Federal Research Station for Agroecology and Agriculture, Evaluation of Ecological Measures,
Reckenholzstrasse 191, CH-8046 Zurich, Switzerland
Institute: Swiss Federal Research Station for Agroecology and Agriculture, Evaluation of
Ecological Measures, Reckenholzstrasse 191, CH-8046 Zurich, Switzerland
Partner 8: Eidgenössische Forschungsanstalt für Agrarökologie und Landbau (FAL,

Switzerland)
SCIENTIFIC TEAM
Name Tel Fax E-mail

Dr. Felix Herzog 41 1 377 71 11 41 1 377 72 01 Felix.Herzog@fal.admin.ch
Dr. Yvonne Reisner 41 1 377 71 11 41 1 377 72 01 Yvonne.Reisner@fal.admin.ch
Joao Palma 41 1 377 71 11 41 1 377 72 01 Joao.Palma@fal.admin.ch
Riccardo De Filippi 41 1 377 71 11 41 1 377 72 01 Riccardo.Defilippi@fal.admin.ch
TIME SPENT ON THE DIFFERENT WORKPACKAGES
Time spent on different workpackages months 36-42
0
Technical annex 2.0 2.0 - - - 1.0 2.0 48.0 5.3 - 60.3
Modified (CMC1) 2.0 5.0 - - - 0.0 2.0 45.0 5.3 - 59.3
Year 1 0.5 2.0 - - - 0.0 0.5 15.7 0.0 - 18.7
Year 2 - 2.0 - - - 0.0 0.5 20.5 0.0 - 23.0
Year 3 - 0.0 - - - 0.0 5.5 17.7 0.0 - 23.2
Felix Herzog - 0.0 - - - 0.0 0.0 0.2 0.0 - 0.2
Yvonne Reisner - 0.0 - - - 0.0 0.0 3.6 0.0 - 3.6
Joao Palma - 0.0 - - - 1.0 1.0 4.0 0.0 - 6.0
Riccardo De Filippi - 0.0 - - - 0.0 0.0 4.5 0.0 - 4.5
Total month 36-42 - 0.0 - - - 1.0 1.0 12.3 0.0 - 14.3
Total (month 1-42) 0.5 4.0 - - - 1.0 7.5 66.2 0.0 - 79.2
Contribution to Workpackages
WP1. Silvoarable modelling strategies (2.0 person-months)
WP1 was completed during the first year.
WP2. European silvoarable knowledge (2.0 person-months)

During the months 36-42 of the project, FAL was not involved in WP2.
WP6. Biophysical integrated plot modelling (1.0 person-month)

FAL helped to calibrate the plot model for the trees and for the crops.

WP7. Economic modelling at the plot scale (2.0 person-month)

FAL submitted weather data (precipitation, temperature, radiation) and the soil
characteristics for each Land Unit in each LTS. These are input-parameters to run the Yield-
sAFe-model (Excel-Version) and calculate the yields for crops and trees. Also the calculation
of the farm size in relation to the Land Units was done by FAL.
FAL contributed to the modelling development and solving technical problems.
João Palma went in December 2004 for 2 weeks to Cranfield University to work on the
models.
WP8. Scaling-up to the farm and the region (48.0 person-months)

T8.1. and T8.2: Selection of landscape test sites (LTS) and acquisition and digitising of
the spatial data needed
1) Work at the level of LTS
In the months 36-42 of the project all work for T8.1. and T8.2 were finished.
The environmental assessment (methodology and some results) at the level of the LTS is still
in progress. At the IALE Congress 2005, a poster about the integration of environmental and
economic assessment at landscape scale will be presented (Palma et al., 2005).
The assessment of soil erosion, nitrate leaching, and carbon sequestration are dependent on
results of YIELD-SAFE and FARM-SAFE. Because there are still some difficulties for
calibration of the biophysical model, the assessment of the environmental indicators will be
finished until July 2005.
To discuss the “Triple Quadrant Approach” to calculate nitrate leaching, Joao Palma visited
Herman van Keulen (Wageningen) in November 2004.
2) Defining target regions for silvoarable agroforestry for Europe (for five tree species
Populus ssp., Pinus pinea, Juglans ssp., Quercus ilex, Prunus avium)
This part of the project is finished. The paper “Target regions for silvoarable Agroforestry in
Europe” was submitted to the Journal “Ecological Engineering” (see Reisner et al., 2005).
WP9. European guidelines for policy implementation (5.3 person-

months)
The regional scenario testing resulting from WP8 will enter the policy analysis and
recommendations elaborated in WP9. Interactive maps at the regional scale (examples) and
at the European scale, which visualise the result of different policy scenarios, will be
developed as a component of the ‘Agroforestry Policy Game’ (Deliverable 9.3).
Meetings, FAL participated:
FAL organised the CMC 6 (6th Consortium Management Committee) in Zurich: Novvember
2004.
FAL organised the “final scientific meeting” in Zurich, November 2004.
FAL visited WU as working visit at the end of November 2004 (to discuss the triple quadrant
approach – for the assessment of nitrate leaching)

FAL visited Silsoe (CRAN) in December 2004 to solve some technical problems in the
modelling and for the calibration of the models to use in the 19 LTS:
SAFE reference list of FAL for the months 36-42 and for the year 2005
De Filippi, R.; Reisner, Y.; Herzog, F. (2005): Data availability and use of GIS to support
agroforestry policies in Europe. 6th Geomatic Week on „high resolution sensors and their
applications“, Conference in Barcelona. 8th February – 11th February 2005.
De Filippi, R.; Reisner, Y.; Herzog, F.; Dupraz, C.; Gavaland, A.; Moreno, G.; Pilbeam, DJ.
(2004): Modelling the potential distribution of agroforestry systems in Europe using GIS. In
18th conference EnviroInfo 2004, Geneva, 21th October – 23th October 2004.
Palma, J,; Bregt, A.; Bunce, R.; De Filippi, R.; Herzog, F.; Van Keulen, H.; Mohren, G.;
Reisner, Y. (2004): Assessing the environmental effects of agroforestry at the landscape
scale. Ecological Engineering. In prep.
Palma, J.; Graves, A.; Bregt, A.; Bunce, R.; Burgess, P.; Garcia, M.; Herzog, F.; Mohren, G.;
Moreno, G.; Reisner, Y.; de Fillippi, R. (2005): Landscape Integration of Economic and
Environmental Indicators to Assess Silvoarable Agroforestry in Spain. European IALE
Congress 2005 on “Landscape Ecology in the Mediterranean: Inside and outside
approaches”. Portugal – Faro, March 29 - April 2, 2005.
Reisner, Y.; Herzog, F. and De Filippi, R. (2004): Target regions for silvoarable Agroforestry
in Europe. Ecological Engineering. submitted.
Reisner, Y.; Palma, J.; Herzog, F. (2004): Assessing the feasibility of silvoarable agroforestry
at different spatial scales. GfÖ Conference, Giessen, Germany, 13.- 17. September 2004.
Future work
Conferences:
European IALE Congress 2005 (International Association for Landscape Ecology) 29 March -
2 April 2005 in Faro, Portugal. Oral presentation. Title: „Landscape Integration of Economic
and Environmental Indicators to Assess Silvo-Arable Agroforestry Options in Spain“ (Palma,
J.; Graves, A.; Bregt, A.; Bunce, R.; Burgess, P.; Garcia, M.; Herzog, F.; Mohren, G.;
Moreno, G.; Reisner, Y.; de Filippi, R.).
6th Geomatic Week on „high resolution sensors and their applications“, Conference in
Barcelona. 8th February – 11th February 2005. Oral presention of Riccardo De Filippi about
„Data availability and use of GIS to support agroforestry policies in Europe“.
Significant difficulties or delays experienced during the reporting

period
The delays in the establishment, parametrisation and quality checks of the biophysical model
made it virtually impossible for us to fulfil our commitments. In an exceptional effort of all
people involved (not only FAL contractor but also the partners and other contractors involved
in WP8 which is co-ordinated by FAL) we managed to provide some results. It will not be
possible, however, to analyse and interpret them as carefully as we should until the project
will be over. Some of this work will be continued beyond the end of the SAFE project, but the
outcome will only be available later.

APCA (Assemblée Permanente des Chambres d’Agriculture) Report
14 Contractor 9: APCA
Contractor 9 : APCA (FRANCE)

Assemblée Permanente des Chambres d’Agriculture
9 avenue George V
75008 Paris, France


Fabien Liagre 33 4 99 61 26 40 33 4 67 52 21 16 liagre@ensam.inra.fr
Thomas Borrell* 33 4 99 61 26 40 33 4 67 52 21 16 borrell@ensam.inra.fr
Pierre Savy 33 1 53 57 11 43 33 1 53 57 10 05 pierre.savy@apca.chambagri.fr
Nathalie Galiri 33 1 53 57 11 33 33 1 53 57 10 05 nathalie.galiri@apca.chambagri.fr
Arnaud Petit 33 153 57 10 66 33 1 53 57 10 05 arnault.petit@apca.chambagri.fr
Thierry Fellmann 33 153 57 11 45 33 1 53 57 10 05 thierry.fellmann@apca.chambagri.fr
* Student
Time spent on the different workpackages during the last period
WP 1 WP 2 WP 7 WP 8 WP 9 Total
Fabien Liagre 0.9 2 1.9 1.2 6.0
Thomas Borrell 1.5 1 2.5
Pierre Savy 0.2 0.3 0.5
Nathalie Galiri 0.2 0.2
Arnaud Petit 0.6 0.6
Thierry Fellmann 0.2 0.2
Total 6 months 0 0.9 3.5 3.1 2.5 10.0
Time spent on the different workpackages during the 42 months
WP 1 WP 2 WP 7 WP 8 WP 9 Total
Technical annex 1 10 9 9 8.4 37.4
Total First year 0.7 3.5 0.9 1.3 0.6 7.0
Total Second year 0.2 5.8 1.7 2.9 2.1 12.7
Total Third year 0 2.5 5.1 3.4 4.6 15.6
Total Fourth year 0 0.9 3.5 3.1 2.5 10.0
Total 42 months 0.9 12.7 11.2 10.7 9.8 44.6
WP2 European silvoarable knowledge (10 person-month)
Improving the database of extant silvoarable systems
Fabien Liagre, with the help of Isabelle Lecomte has finished the database of extant
silvoarable systems during this last period. This database, written in Access software, helps
the user to locate all the traditional and modern silvoarable systems in the different countries
of the Programme. Each site, identified by specific data and photo, is defined by a range of
maps which help to locate the area. (see WP2 report for more information).

Study of farmers’ reaction about agroforestry
During the last 6 months, Fabien Liagre made the whole analysis of the data from each
partner. He finished the deliverable 2.3 (ex 8.3).
In France, all the results from the interviews have been presented in each target region.
Three local meetings have been organised in Prahecq (Poitou-Charentes), Orleans (Centre)
in December 04 and in February 05 in Besançon (Franche Comté). These meetings were the
opportunity to discuss all the results with the technicians who helped to create the sample
and the farmers who were interviewed.
The results we observed for France were quite surprising. In very productive region (Centre)
or on the contrary, in region where it’s difficult to maintain the agricultural area face to the
forest development (Franche Comté), the reaction of the farmers was very enthusiastic to the
silvoarable idea (see Table 125). If many farmers pointed out all the technical difficulties,
they showed a deep interest to be informed about the potential of these systems – 80%
wanted to be contacted again.
60%
50%
Number of Farmers (%)
40%
30%
20%
10%
0%
Centre Franche Comté Poitou- Total
Charentes
Enthusiastic Interested Undecided Against
Table 125: Are the French farmers interested in creating some silvoarable project? A
third of them said that it could be an option in a short term future. And 12 % seemed to
be really motivated.
Even if the percentage of interested farmers in creating some silvoarable projects is less
important than in the European study, almost 30% of the French farmers said they would
think about it. In the Region Centre, we found a large number of farmers against this
possibility. On the contrary, we have more possibilities to find an interested farmer in Poitou-
Charentes.
This result was a strong surprise for the Chambers of Agriculture. We realised an opinion poll
with all the persons of the Agriculture Chambers who worked in the Farmer’s reaction study.
Almost 50 questionnaires were sent to them to sound up their opinion about the farmers’
reaction. And we received 25 answers. In conclusion, a few technicians were able to forecast
these results.

60
50
40
% Answers
30
20
10
Interested % Undecided % Against %
Technicians Forecast
Farmers AnswersRéponses Agriculteurs
Table 126: Technicians forecast about the answer of the farmers according to their
interest in creating or not a silvoarable project. If the technicians share the same
interest as the farmers concerning the possibilities of development of agroforestry,
they nonetheless think that this option wouldn’t interest so much the farmers.
Before each regional meeting, we phoned to each farmer, two years after the interviews, to
sound up again if they were thinking about a silvoarable project, if they had changed their
mind about agroforestry (see Table 127).
60%
50%
40%
Interested
30% Undecided
Against
20%
10%
0%
Interview Phone call
Table 127: Evolution of the interest in creating a project in the farmers who have been
interviewed in 2003.
After 2 years, the farmers interested in creating a project are less important, although we
have still 10 farmers who are still enthusiastic. Finally, only half of the interviewed don’t
imagine to plant trees in their cropping area, although some of them don’t close the door to
this eventuality… In case of great subsidy, they can adopt their farming system!
We must also underline that 7 farmers have initiated some discussions with their technicians
to see how to set up their silvoarable plot for 2005 (2 in Poitou-Charentes, 2 in Centre and 3
in Franche Comté)! From these 7 farmers, 5 had said that they were interested, one that he

was undecided and the last one was against any project… This last farmer changed his mind
for some environmental goal. In fact, his village decided to protect their water catchment’s
area and to adopt some agro-environmental measures to maintain a good water quality. He
proposed therefore himself a silvoarable measure. This programme will concern numerous
farmers of this area. We can add that, again in this environmental approach, another farmer
we interviewed could propose an agroforestry measure for the same reason in his village.
The rest of the farmers who want to plant trees in their crop give some economical reasons
(diversification and inheritance) or just want to improve the landscape.
One of them told us during the interviews: “Your system is fantastic! Sorry to tell you that, but
if you don’t succeed to change the European regulations to take into account the agroforestry
specifications, you are an idiot! Sorry to tell you that!!”
To learn more about some interesting results about the WP2 interviews, see the WP2
report.
To know which kind of actions APCA took to avoid to be considered as an “idiots”

institute, see the contributions of APCA to WP9…
WP7 Economic modelling at the plot scale (9.0 person-months)

During the fourth year, Thomas Borrell and Fabien Liagre, in collaboration with Christian
Dupraz - INRA, have realised all the technical and economical simulations for the French
context. APCA participated to the redaction of the French chapter for the deliverable 7.2 -
Plot economics of European silvoarable systems report – leaded by the partner Chavet.
APCA was also in charge of the redaction of the deliverable 7.3 - Optimum AF systems for
different regions report - in collaboration with Christian Dupraz.
Face to the late we observed in the results we should have received from Yield-sAFe, it was
decided to go on our simulations using the Ler-Safe data to feed Farm-sAFe.
All the results of our simulations have been presented to all the Chambers of Agriculture
which have provided the economical data in the 3 regional meetings we named before.
Main objectives
The main objectives for APCA were:
• To provide economic data on representative farms of targeted areas for silvoarable

dissemination in 3 areas of France, using existing national farm survey information
(ROSACE network)
• To predict outputs at plot and farm scale of three different target French areas
including a totally non-forested intensive cereal-growing area in the Paris basin
• To participate in the elaboration of economic model to realize the different scenarios

(Farm-sAFe, Yield-sAFe and Ler-Safe)
Data references and main hypothesis
The forestry references

The revolution duration, timber production and production techniques (initial density,
prunings, thinnings, sward maintenance, and final density) were determined by expert
knowledge, in accordance with available documentation.

The densities correspond to the schedule of conditions of the French circular “Forêts de
production” and to forestry organisms’ advises.
Individual density Revolution duration Mean annual production

piece of (trees/ha) Volume (years) (m3/ha/year)
timber at at felling Good Land Bad Good Land
Bad
felling initial final (m3/ha) land unit land land unit
land unit
(m3/tree) unit medium unit unit medium
Walnut 1 200 100 100 46 53 60 2,17 1,89 1,67
Wild cherry 0,8 800 150 120 50 55 60 2,40 2,18 2
Poplar 1,5 200 200 300 19 22 25 15,79 13,64 12
Table 1: Densities, revolution duration and mean annual and total productions for
walnut, wild cherry and poplar. With walnut, 2 thinnings of 50 trees/ha are realised at
1/3 and 2/3 of the revolution ; with wild cherry, 3 thinnings are realised at 1/3 (400
trees/ha), half (200 trees/ha) and 2/3 (50 trees/ha) of the revolution.
Supports for afforestation on agricultural land vary in function of the region and of the tree
species: as the poplar revolution is shorter, the PCPR is available for 7 years instead of 10.
Type of farm Ly-Lc Hy-Hc

Hy-Lc
(region) (Poitou- (Franche-
(Centre)
Charentes) Comté)
Walnut and wild cherry
Poplar
Table 2 : Regional supports for afforestation on agricultural land for walnut, wild
cherry and poplar (year 2003). The PCPR is the Compensation Payment for the loss of
agricultural income.
Franche-Comté is a particular region. More than 50% of the area are already woodlands,
thus afforestation is not encouraged: there is no support available for new forestry plantation.
Everywhere in France, newly afforested plots benefit from an exemption from land tax: for
10 years with poplar, 50 years with walnut and wild cherry. In our simulations, this land tax is
comprised between 30 €/ha (Centre) and 39 €/ha (Poitou-Charentes).
Reference data in agriculture

All arable data come from the Farm observatory ROSACE. Thanks to this typology of farms
made by the regional Chamber of Agriculture, several types of farms are defined and
described, each one corresponding to the mean of 5 to 10 farms selected by the Chambers
experts. Each year, the economical inputs are re-calculated (yield, net margin, farm costs,
labour and CAP payment). In addition, all the technical orientations and strategies of the farm
are described.
We selected 3 types of farm, which we shall now designate with 4 initials:
• Hy-Lc: High yields and Low fixed costs
• Hy-Hc: High yields but High fixed costs
• Ly-Lc: Low yields and Low fixed costs
For each of them, the ROSACE typology indicates:

• The cropping area of the farm, distinguishing tenant farming and property;
• The crop rotation in function of the quality of the soil (up to 3 Land Units: best,
medium, worst);
• The mean yields, attributed to the medium Land Unit (for the best and worst Land
Units, we respectively assumed an increase and a decrease of 10% of the mean
yields);
• The variable costs, assignable fixed costs and fixed costs and labour.
• The prices of the products and sub-products (straws of the wheat) and the CAP
payments of the farm Single Farm Payment, SFP).
The selection of each type of farms, various partners from the Chambers of Agriculture have
participated: Camille Laborie, who is in charge of ROSACE in APCA, Anne-Marie Meudre
(Franche Comté), Catherine Micheluzzi (Poitou-Charentes) and Benoît Tassin (Centre).
Cropping area of
the farm (ha)
Total
Fixed Mean Gross Net
Typical area Net
costs Crop yield Marg Marg
rotation(s) (ha) Margin
property
(€/farm) (t/ha) (€/ha) (€/ha)

farming
(€/farm)
tenant
Total
(a) wheat 8
oilseed 2 t/ha)
72 8 80 36805 18045
or oilseed 4 918 711 30,0
Hy-Lc
(b) wheat set

oilseed – 323 116 8,0
aside
6,5
wheat 2 t/ha)
sunflower oilseed 3,2 674 422 14,1
56,4 37,6 94 28785 17153
wheat sunflow
oilseed 2,5 799 547 28,2
er
Ly-Lc
sunflower set
– 318 66 9,4
aside
(a) wheat 6,7
wheat 2 t/ha)
wheat oilseed 3,5 728 413 14,1
97,5 32,5 130 40370 wheat maize 7,5 555 240 28,2 12031
maize
or
set
(b) wheat – 313 -2 9,4
Hy-Hc
wheat aside
oilseed
Table 3 : Main economic data and total net margin (€/farm) for every type of farm.
Rotation (a) corresponds to the best land units, rotation (b) to the worst. Set aside is
realised on 10% of the total farm area.

The Net Margin is equal to the Gross Margin minus the fixed assignable costs (land tax and
machinery costs). The Total Net Margin is equal to the Net Margin minus the fixed costs (rent
of land, amortisation and maintenance of the buildings, social contributions, banking costs).
Labour costs are not taken into account.
The profitability threshold yield

With the development of the trees, the crop yield decreases progressively. Below a certain
level, the crop is not more profitable, above al near the tree area. For each crop of the three
types of farm, the threshold yield was first determined according to the price of the product,
the CAP payment and the variable costs, assignable fixed costs and a part of the fixed
costs14. As the results, in proportion of the mean yield of each crop, were roughly the same in
the three farms, we fixed this proportion in order to facilitate the extrapolation to other types
of farm.
Mean yield Profitability Mean yield in Profitability threshold

crop
in the farm threshold yield Hy-Hc yield in Hy-Hc
Winter wheat 100 % 50 % 6,7 t/ha 3,35 t/ha
Maize 100 % 70 % 7,5 t/ha 5,25 t/ha
Oilseed rape 100 % 60 % 3,5 t/ha 2,1 t/ha
Table 4: Profitability threshold yield in proportion of the mean yield in the farm and
example for the farm Hy-Hc
The threshold yield is the same in every farm, whatever the land unit is. Thus it shall be
reached more quickly in the worst land unit than in the best land unit.
Main management features of the agroforestry systems

For each type of farm, we simulated the introduction of 2 agroforestry designs in the 3 land
units (best, medium, and worst):
• Plantation at 50 trees/ha, on 40 m spaced tree-lines;
• Plantation at 120 trees/ha, on 22 m spaced tree-lines.
The tree strip is 2 m wide. The width of the intercropped alley is respectively of 38 m and 20
m, thus the maximum crop area represents 95% of the initial area at 50 trees/ha and 91% at
113 trees/ha.
With walnut and wild cherry, an early thinning is realised when the timber volume reaches
0,1 m3 (around the years 10-13), therefore the final densities are different from the poplars’
one (see Table 5).
14
If the crop is abandoned on a part of the cropping area, we assume that the fixed costs should
decrease a little; thus they must be taken into account in the calculation of this threshold yield.

Agroforestry Forestry
Tree
Density Timber Density Timber Tree RA-
Productio Productio RA-
(trees/ha) volume (trees/ha) volume products
(m3/ha) (m3/ha) biomass
initial final (m3/tree) initial final (m3/tree)
50 40 1,2 48 0,36 0,48
Walnut 200 100 1 100
120 80 1,08 86 0,66 0,86
Wild 50 40 0,96 38 0,22 0,32
800 150 0,8 120
cherry 120 80 0,92 74 0,42 0,62
50 50 1,8 90 0,3 0,3
Poplar 200 200 1,5 300
120 113 1,76 199 0,66 0,66
Table 5: Initial and final densities, volume of an individual piece of timber and
production in forestry and in the simulated agroforestry systems; tree Relative Area
(RA)-biomass and tree RA-products
The crops Relative Areas (RA) have been fixed for 3 hypothesis: optimistic, probable and
pessimistic.
The pessimistic hypothesis means that the LER-biomass is equal to 1. Therefore, the crop
RA is equal to: (1 – tree RA-biomass).
The optimistic crop RA is determined according to 2 constraints:
• The crop RA must be inferior to the maximum intercropping area
• We also assumed to fix a ceiling for the LER-biomass of 1.4. Thus the crop RA is
equal to: (1.4 – tree RA-biomass). This ceiling of 1.4 was reached with walnut and
poplar at 120 trees/ha, so the crop RA seems quite low with regards to the maximum
intercropping area.
We assumed a probable crop RA as the arithmetic average of the 2 previous values

(pessimistic and optimistic) (see Table 6 and Table 7).
Width
Initial Width of the Maximum
between Pessimistic Probable Optimistic
density intercropped intercropping
tree lines crop RA crop RA crop RA
(trees/ha) alley (m) area
(m)
50 40 38 0,95 0,64 0,79 0,94
Walnut
120 22 20 0,91 0,34 0,54 0,74
Wild 50 40 38 0,95 0,78 0,85 0,93
cherry 120 22 20 0,91 0,57 0,72 0,87
50 40 38 0,95 0,7 0,8 0,9
Poplar
120 22 20 0,91 0,34 0,54 0,74
Table 6: Crop RA in function of the tree species, density and optimism level. Bold
values are those which depend on the ceiling of 1.4 for the LER-biomass.

Width Width of LER-biomass reached with LER-products reached with

Initial between the the the
density tree intercrop
(trees/ha) lines Pessim. probabl Optimist Pessim. probabl Optimist
ped alley
crop RA crop RA crop RA crop RA crop RA crop RA
(m) (m)
Walnut 50 40 38 1 1,15 1,3 1,12 1,27 1,42
120 22 20 1 1,2 1,4 1,2 1,4 1,6
Wild 50 40 38 1 1,07 1,15 1,1 1,17 1,25
cherry 120 22 20 1 1,15 1,3 1,19 1,34 1,49
Poplar 50 40 38 1 1,1 1,2 1 1,1 1,2
120 22 20 1 1,2 1,4 1 1,2 1,4
Table 7 : LER-biomass and LER-products in function of the tree species, density and
hypothesis of optimism for the intercrop Bold values are those which depend on the
ceiling of 1.4 for the LER-biomass.
Economic hypothesis
CAP payments
In agriculture, the crops area benefits from the Single Farm Payment (SFP): it was calculated
on the basis of the historical references of each farm, in accordance with the way France
decided to implement the new CAP in 2006.
In the basic scenario, we assumed that the intercrops are eligible to the SFP proportionally to
the area of the plot that they occupy. It is the present situation in France. The rights
corresponding to the tree area could be transferable to another eligible area which doesn’t
benefit from a payment right. In our simulations, we did not attribute them to new plots,
considering therefore that these rights were lost for the farmer.
Tree grants
In our basic scenario, agroforest trees benefit from the same establishment payments as the
forest trees: 50% of the costs of the 4 first years in Poitou-Charentes and Centre. It
corresponds to the present situation, permitted by the circular “Forêts de protection” which
relies on the line i of the French National Rural Development Programme. However an
agroforest plot cannot benefit from neither the PCPR nor the exemption of land tax.
In France, an agro-environmental measure called “agroforest habitats” can be contracted

under certain conditions, but it still faces administrative difficulties and is not available in most
of the departments, thus it was not taken into account in our simulations.
Costs and prices

Some key points have to be underlined:
• The cost of sward maintenance is higher in forestry than in agroforestry. In forestry, at

the beginning of the revolution, sward maintenance is realised thanks to two grindings
instead of one for the maintenance of the tree strip in agroforestry.
• The farmer makes all operations himself, except the marking out and plantation of the
young trees. Both of these operations are charged 15 €/h. The timber prices
correspond to standing trees, thus neither the harvesting cost is taken into account.
• In a cash flow approach, the basic scenario doesn’t include the labour cost for the
farmer. While in a farming management scenario, we consider an hourly cost of 7,62

€/h (minimum salary in France). In this last approach, it’s therefore possible to
evaluate the efficiency of the farmer labour.
As it seems impossible to anticipate the future evolution of prices and costs, we assumed
constant values. For instance, a rise or a drop of timber value would respectively increase or
decrease the tree revenue.
Main results
Labour impact for one silvoarable hectare

temps de travail temps de travail
(h/ha/an) (h/ha/an)
16 16
agriculture agriculture
14 cultures intercalaires 14 cultures intercalaires
arbres arbres
12 12
10 10
8 8
6 6
4 4
2 2
0 0
1 6 11 16 21 26 31 36 41 46 1 6 11 16 21 26 31 36 41 46
année année
Case 1: Plantation of 120 trees/ha Case 2: Plantation of 50 trees/ha
Table 128: Labour evolution in the management of a silvoarable plot during the tree
rotation, separating the crop from the tree labour.
An essential condition for adopting agroforestry from the farmers’ point of view is that they
don’t want to devote more time to a new system. If the farmer planted more trees (case 1),
he would need 1 to 1.5 days each year to maintain the trees. But in the second half of the
rotation, the labour decreases progressively due to the fact that trees don’t need more
special maintenance and that the intercrop activity is reduced. If he plants fewer trees, the
impact during the first years is poor. With the small density, the intercrop activity is longer,
because the crop yield is not so affected by the trees. The labour requested in the second
half of the rotation is therefore lower but very near from the initial scenario.
Prediction of yield evolution
Crop yield evolution

Predicting the crop yield during the second half of the rotation is a perilous venture. If we
know the behaviour of the intercrop during the first half thanks to experimental measures on
existing plots, we asked the bio-economics model to predict the yield evolution. In our
simulation, as we said, we used the LER-Safe prediction. We made the essential hypothesis
that the LER must be include between 1 and 1.4. This condition helps us to determine a
possible range of crop yield evolution, from the pessimist one to the optimist one (see Table
129).

Tree plantation Tree Harvest
100 100
Optimist
80 80
Crop yield (%)
60 60
Pessimist
40 40
Intercrop Yield
20 Pure Crop 20
0 0
0 Time
Table 129: Evolution of the relative intercrop yield according to optimist or pessimist
view about the tree competition. Case of one ha of wild cherry with an initial density of
120 trees/ha for a final density of 80 trees/ha.
In this example of a plantation of wild cherry at 80 trees/ha (final density), which means a
distance between the trees rows of 25m, the crop yield represent more than 90% during the
first half of the tree rotation. According to the interaction level, the crop yield varies between
30 and 75 % of the pure crop yield of reference the year before harvesting the trees.
The crop yield depends on different parameters:
• The parameters due to some initial choices: the crop nature (a sunflower will be more
affected by the shadow of the adult trees than a cereal), the density of the plantation
and the distance between the lines, choice of the land unit (a deeper soil will be more
adapted),...
• The parameters depending on the capacity of the farmer: well pruned trees, tree root
maintenance (root cutting), …
In our economical scenarios, we have tested the different level of interaction.
Tree yield evolution

As for the crop yield estimation, we put forward the hypothesis of different level of timber
productivity. But for our simulations, we only use one prediction of timber production. To
validate our approach, we use a very cautious estimation of production (see Table 130). Our
results can therefore be considered as the minimum result we can get from our hypothesis.

140 140
Interval
120 120
Basic
Standing volume (m3/ ha)

100 scenario 100
Pure Optimist
80 plantation 80
60 60
40 40
Pessimist
20 20
0 0
5 10 15 20 25 30 35 40 45
Time
Table 130: Range of timber volume evolution for an initial plantation of 120 wild
cherry. The figure indicates of the cautious hypothesis of standing volume we used
for our simulations (77 m3 for 80 final trees).
Cash flow impact

To evaluate the impact of the project on the cash flow, we must distinguish first the
investment cost and then the evolution of the annual cash flow depending of the crop yield
evolution and the possible over cost to crop between the trees in comparison with a pure
crop system.
Initial investment
The poor number of trees to plant in an agroforestry system reduces considerably the
investment cost if we compare with a current afforestation cost on agricultural land. The tree
cost is nonetheless higher. The owner will choose a better quality of the trees and will have
to protect each one with a strong protection: each tree has a possible future value and
demands a special attention.
The total cost of a plantation (without subsidy) varies between 500 and 1000 euros/ha
according to the tree specie (the walnut plantation being the most expensive). This cost
represents between 20 to 60 % of the average cost in the case of common land afforestation
(see Table 131).
Afforestation
1 233 €/ha
120 trees/ha
Poplar 695 €/ha
367 €/ha
50 trees/ha
1 518 €/ha

267 €/ha
1 633 €/ha
Walnut 1 034 €/ha

517 €/ha
Table 131: Comparaison of the investment in agroforestry and forestry scenario,

WITHOUT subsidy.

In France, it’s current to get a subsidy of 40 to 70% to cover the investment cost and the
maintenance cost of the trees during the 4 first years (except in Franche Comté).
Since 2004, the French Government decided to suspend all economic aids to the land
afforestation, excepted for agroforestry. In our simulations, we decided to conserve this aid,
to be able to compare between the two options (see Table 132).
Afforestation
617 €/ha
120 trees/ha
Poplar 348 €/ha
184 €/ha
50 trees/ha
759 €/ha

134 €/ha
817 €/ha
Walnut 517 €/ha

259 €/ha
Table 132: Comparison of the investment in agroforestry and forestry scenario, WITH
subsidy.
Cash flow evolution

Evolution of the cash flow at the plot scale
The cash flow evolution will depend of the crop yield evolution and the LER level we have
selected and the final density. For example, in the Table 133, we’ve illustrated the cash
evolution for two different densities but for a medium LER level.
100 100
90 50 trees/ha 90
80 to 90 %
80 70 to 85 % 80
70 70
% Annual Gross Margin
120 trees/ha
60 60
50 30 to 60 % 50
40 40
30 30
20 Silvoarable Gross Margin 20

Agricultural Gross Margin
10 10
0 0
Time Trees Harvesting
Plantation
Table 133: Evolution of the annual cash flow for a probable scenario with wild cherry
(LER=1.07 for a density of 50 trees/ha and 1.15 for a density of 120 trees).

Being cautious in our forecast, we notice nonetheless that at half of the rotation, the gross
margin still represent 80 to 90 % of the agricultural gross margin. We must underline that in
our simulations, we’ve considered that the crop payment area is reduced progressively by
the tree area. In case of the silvoarable area was eligible in its totality, the impact on the cash
would be sensible, above all in some regions where man get poor crop yield and where the
crop payment is essential in the gross margin calculation (Franche Comté for example).
Let’s also underline the fact that in the INRA experimental plots, the LER reaches more 1.3
than 1.15 that we have chosen in our simulation with an initial density of 120 trees/ha.
Evolution of the cash flow at the farm scale
At the farm scale, one of the first questions of the farmer is about the importance of the area
to plant. Does he have to plant on a big area? In several plots or in a single plot? There is no
only one answer. According to the strategy of the farmer, a large range of scenarios is
available. The choice will depend to the cash flow context and to know if the farmer can
support a strong investment or not, and above all if he aims to decrease progressively his
crop activity or not. The labour availability is also a strong parameter to decide which area to
plant. According to our simulation and experimental experience, we often recommend not
planting more than 10 % of the cropping area. In that case, the impact on the farm gross
margin is less than 3 % in average on the first half of the tree rotation. A gradual plantation
will allow a reduction of the cash flow impact (see Table 134).
435 %
191% 183%
178% 180%
175 Farm with 8% silvoarable area 175 Farm with 8% silvoarable area
Farm with 100 % of cropping area Farm with 100 % of cropping area
150 150
125 125
100 100
75 75
50 50
0% 20% 40% 60% 80% 100% 120% 0% 20% 40% 60% 80% 100% 120%
% of the tree rotation % of the first tree rotation
a. Case of a single plantation b. Case of a gradual plantation
Table 134: Comparison of the cash flow evolution when the farmer plants 8 % of his
cropping area (50/50 Walnut/Wild cherry). We compare the option where the farmer
would plant the silvoarable area in once time or if he decides to plant 2 % every 5
years during 20 years.
A gradual plantation will also allow a soft distribution of the timber income in the time from
the moment where the owner begins to harvest the first mature trees (case b). From this
moment, the timber income is regular. In our example, he can harvest the trees every 5
years. In this context, the farm gross margin increase by 15 %. According to the importance
of the plantation and of the species he planted, a farmer could increase his farm income
between 10 to 100%. Of course, it can suppose a long term to wait for the farmer before the
first tree harvest…
Profitability of a silvoarable investment
Comparing a silvoarable scenario with agricultural scenario

For our simulations, we have selected 3 kinds of farms:
• Farm with good crop yields and few fixed costs.

• Farm with medium crop yields with few fixed costs.
• Farm with medium crop yields and high fixed costs.
For each farm, corresponding to each region of the LTS of the WP8, we have run different
scenarios according to:
• the tree density: 120 versus 50 for the initial density which corresponds to a final
density of 80/40.
• the LER level: optimist, probable and pessimist
• the land unit: good/medium
• the 3 tree species: poplar, walnut and wild cherry
108 scenarios have been run in total (36 scenarios / LTS). The Table 135 shows a synthesis
of the Agricultural Values for all these scenarios we have calculated for each specie
according to the level of LER.
Walnut Wild Cherry Poplar

100%
Agricultural
% of realised simulations
80% Value Index

> 1,35
60% 1,20 - 1,35
1,05 - 1,20
40%
0,95 - 1,04
20% < 0,95
0%
le
le
le
pr st
pr st
pr st
ist
ist
ist
ab
ab
ab
i
i
tim
tim
tim
im
im
im
ob
ob
ob
ss
ss
ss
op
op
op
pe
pe
pe
Scenario for intercrop productivity
Table 135: Profitability of the silvoarable scenarios according to the tree specie and
the LER level.
A first interesting result is that the silvoarable scenarios are at least as profitable as the
agricultural scenario.
Walnut timber is actually the most expensive timber on the market. For a same duration of
rotation, the best results have been logically obtained with the walnut than the wild cherry.
The period of harvesting time is a key parameter in the profitability calculation (see Table
136).

2,00
1,80
1,60
1,40
1,20
1,00
0,80
0,60
0,40
0,20
0,00
Very
Trèswellbien
pruned Bien
Well pruned
formé Badly
Mal pruned
formé
formé (50 ans) (60 ans)
40 years 50 years 60 years
(40 ans)
Table 136: Influence of the maintenance quality on the profitablity.

A late in the pruning dates can put the harvesting date back by 10 or 20 years, above all for
some sensitive specie such as the hybrid walnut. In this example, a late of 20 years means a
reduction of 60% of the profitability in comparison of the agricultural profitability.
Comparing a silvoarable scenario with a forestry scenario

We compare also the case where the farmer was hesitating between a forestry investment
rather than a silvoarable investment from a profitability point of view (see Table 137).
Agricultural Value Index

1,50
of a silvoarable scenario
of a pure plantation scenario
1,00
1,55
1,21
0,50 1,04 1,00
0,89
0,48
0,00
Poplar Walnut Wild Cherry
Table 137: Comparison of the profitability of the silvoarable and afforestation scenario
with the agricultural scenario. Silvoarable plantation of 120 wild cherry by ha
characterized by a LER of 1,15.
In this example, we explore the case of a probable LER of 1.15 in the silvoarable option. In
almost all our simulations, the silvoarable options are more profitable than the forestry option.
The forestry option may be more profitable in the case where the crop margin is very poor,
above all if it’s possible to plant some valuable species such as walnut for example.

It’s also interesting to notice that for the poplar, the silvoarable option could be a possibility to
stimulate the poplar market. In France, the poplar area is currently decreasing because of
the price fall of the timber (less than 45 €/m3). Agroforestry could therefore be a possible
strategy to reduce the market risks.
Property holdings evaluation in agroforestry

According to his age, a land owner who plants trees, will not necessary benefit from the
harvest… But, as a farmer told us, a farmer has three possibilities of income: the sale of his
products, the stock variation and the possibility to make a capital gain. In this last option, a
silvoarable plot is a capital which could be evaluated if necessary (inheritance, expropriation,
etc). The land evaluation in agroforestry is the combination of the agricultural land evaluation
and the future value of the trees (see Table 138).
16 000 €
No commercial value with commercial value
14 000 €
12 000 €
10 000 €
Euros by ha
8 000 €
6 000 €
4 000 €
2 000 €
0€
10 20 30 40
Age of the trees (years)
Agriculture agroforestry
Table 138: Evolution of the monetary value of the silvoarable land according to the
age of the trees. In agroforestry, this value is the sum of the agricultural value plus the
timber future value. If the young trees could have a future value, for example at 10
years old, they don’t necessary have a commercial value in the sense that the
landowner can not expect some income if he cut them.
In this example of a wild cherry plantation, the capital evaluation may represent between
twice and four time the agricultural land value according to the age of the trees. In the case
of a walnut plantation, it may represent till 7 times this value 10 years before the tree
harvesting.

Photo 5: In this plot of 4 ha, the wild cherries are 30 years old. The value of the
standing volume is estimated to 4 000 €/ha, which represents the same value of the
agricultural land. But the future value of this plantation is much higher and overpass
the 10 000 €/ha.
Main conclusions
To invest in agroforestry represents a light investment in money and labour comparing with
some new systems of diversification. In our simulations, the profitability reaches 10 to 50 %
with walnut, and -5 to +15 % with wild cherry and poplar, comparing with the agricultural
scenario.
A regular calendar of plantation on a few surfaces is a good option for the farmer (labour and
cash flow impact). 10 % represents between 2 and 3 % of reduction of the farm gross
margin. But in the balanced period, the income increase of more than 15% (mixed plantation
of walnut and wild cherry trees). The gross margin could double if the farmer plants
progressively his whole cropping area. But in that case, it means a stronger impact on the
initial cash flow and demands a consequent labour...
If the best bio-physical option is to plant between 90 to 120 trees by hectare, the best
economical option is to plant a lower density around 70 to 90 trees by hectare. This means a
distance between the trees lines varying between 24 to 36 m.
For more information, see Deliverable 7.2.
Dupraz, C., Liagre, F. & Borrell, T. (2005) The Land Equivalent Ratio of a silvoarable
agroforestry system. In preparation.
Borrell, T. (2004) De l’importance des interactions arbres-cultures sur les performances

économiques de l’agroforesterie tempérée. Mémoire de Diplôme d’Agronomie Approfondie,
ENSAM-INRA, Montpellier. 98 p + annexes
WP8 Scaling-up to the farm and the region (14.0 person-months)

APCA contribute to the realisation of the deliverable 8.1 in collaboration with Anil Graves and
Joao Palma. APCA delivered all the data to A. Graves and J. Palma for their simulations.
APCA spent also a large time to evaluate all the results obtained by Yield-sAFe for the
French LTS, contributing by this task to the calibration of the model Yield-sAFe.

For more information, see Deliverable 8.2.
Graves, A.R., Burgess, P.J., Liagre, F., Dupraz, C. & Terreaux, J.-P. (in preparation) The
development of an economic model of arable, agroforestry and forestry systems. To be
published soon in Agroforestry Systems.
WP9 European guidelines for policy implementation (8.4 person-months)

French forestry and agricultural policies have been scrutinised for bottlenecks on
agroforestry implementation, and possible conflicting rules between forestry and agricultural
policy. This task was divided in two parts:
• Agroforestry and CAP policy
• Agroforestry and land status
In each case, APCA has created some working groups with different partners from the
agricultural and forestry field (administration and professional). Different solutions have been
suggested.
APCA has contributed to the design of a common framework for the implementation of a
European agroforestry scheme based on the data from the models. APCA has organised 4
end-user conferences at the end of the project in France. APCA has also contributed to the
organisation of the Brussels conference and participated to the Madrid end users
conference.
APCA spent a large time in the dissemination of the SAFE results in three different
audiences: the Chambers of Agriculture, the French government and the farm and forest
end-users.
Agroforestry and CAP policy
APCA made lobbying to support agroforestry in the CAP policy.
At the European level, APCA defended the article 41 of the draft Regulation of Rural
Development as member of the Committee of Professional Agricultural Organisations in the
European Union (COPA) and as councillor of the Social and Economical Committee (CSE).
Arnaud Petit manage this lobbying action in the name of APCA to support the idea to finance
the agroforestry plantation on agricultural land during the draft project evaluation by the CSE
and COPA.
At the national scale, APCA has defended the eligibility of the silvoarable plot in the CAP
application. In the existent regulation, when a farmer plants a new silvoarable plantation, he
must deduce the tree area from the eligible area of the plot. Luc Guyau, APCA president,
sent a letter to the Agricultural Minister to demand a total eligibility of the silvoarable area
(see Annex 1). APCA proposed 2 conditions to get this total eligibility:
• More than 50% of the plot should be devoted to an agriculture production
• The tree density must be under 200 trees by hectare
This demand is now discussing in the Ministry of Agriculture. A future meeting with the
Ministry and Professional Organisation is planned to discuss about the place of the isolated
trees in the next policy. (See Annex )

Suggestions of a land status for the agroforestry plot
APCA initiated during the SAFE programme and above all in this last period a discussion
about the definition of the land status in agroforestry. The French Ministry of Agriculture
made a first draft inventory of the different silvoarable systems existing in France and
proposed different solutions to be discussed. These solutions would be explored by APCA
and its partners to make some proposals in the following months to be discussed with the
Government and the Parliament. The objective is to define an official status to be included in
the future Law of Agricultural Direction in 2006.
Photo 6: In Vézénobres, the silvoarable status is a combination of the agricultural rate

and forestry rate, according to each component area.
In Annex , after resuming the different cases existing in agroforestry and the proposals of the
French Ministry, APCA presents the principles way of solutions.
Dissemination of the SAFE results
4 End-user conferences have been hold at the end of the programme, 3 in each region which
has participated in the programme and one national conference in Paris in APCA. These
conferences have been a success.
Luc Guyau, president of APCA, has leaded the Christian Dupraz, Fabien Liagre and Thomas
national meeting in APCA (in the centre, white Borrell have presented the main results of the
shirt). One hundred persons coming from all SAFE programme in each Regional Chamber of
France have attended the meeting. Agriculture.
Photo 7: End-user conferences in France

In each regional meeting, we invited all the technicians and farmers who have been
approached for the interviews inside the WP2, and have contributed to give economical data

for the WP7 and 8. These meetings were also the opportunity to debate about the future of
agroforestry in each region and to imagine how to go on the research development activities.
Each of the 3 provinces has shown a strong interest to carry on this way.
In Paris, on January 26th 2005, has taken place the national conference. Each Regional
Chambers of agriculture and forestry institutes have been invited. APCA invited also the
members of the different Ministry of Agriculture and Ecology. In conclusion of the conference,
strong contacts with administration and professional have been taken to think about how to
defend agroforestry in the new policies (CAP and status).
APCA also decided to organise a national training courses for October 2005 (5th to 7th) to
train the extension services to the management of silvoarable projects.
A special Agroforestry edition of the review “Revue Des Chambres d’Agriculture” will be
published at the end of 2005. This edition will present the main SAFE results. Different
partners of the SAFE programme will participate to this edition.
French prospect of the agroforestry development in a near future
Evolution of the number of silvoarable projects

During the last period of the SAFE Programme, APCA carried on a strong support the
silvoarable projects in France. Many technicians have been involved to supervise the setting
up of projects.
The last estimations in France show that the silvoarable area of recent projects will overpass
the 1200 ha in 2006 (see Table 139), which will concern more than 150 owners (see Table
140). If we include the older projects, the whole area will approach 1 700 ha for more than
170 owners.
1400
1200
1000
800
hectare
600
400
200
0
2000 2001 2002 2003 2004 2005 2006
Year
Cumulated Existing In study First contact
Table 139: Evolution of silvoarable area during the last 5 years in France (only recent
projects are taken into account)

180
160 154
140
Number of actors
120
100
80
80
60
9
40
7
20 2
0 0
0
2000 2001 2002 2003 2004 2005 2006
Cumulated Existing In study Contact
Table 140: Possible evolution of the silvoarable projects number in France for the next
two years.
Thanks to the policy reform in France and to the investment of APCA during these last three
years, we assist to a development of agroforestry in France. This development is very
encouraging for those who carried out this topic. We didn’t count today the number of phone
calls from farmers, land owners or technicians interested to know more about agroforestry.
More than 20 articles in professional reviews or newspaper have been edited during the
SAFE period. 5 farmers’ organisations have asked for a specific training in 2004.
But this success of agroforestry in France supposes to organise its development. Each
project holder must receive right technical and practical information to set up his project in
good conditions. On the other hand, the diversity of goals underlined by these project holders
induces some new questions to the Research Development which induces also more means
for the Research teams.
After the SAFE project…
Research Development Programme

Different partners from the Research Development have proposed a national programme to
organise the development of Agroforestry in France. 17 French Départements participated to
this proposal. The goals of the programme would be:
• To create a national end-user association
• To create a network of demonstration plots in each province
• To create a research development unit, associating INRA, professional organisms

and education institutes.
• To give all kind of information to each silvoarable actor in France

APCA is quite optimistic to receive the agreement

from the French Rural Development Agency to set up
this programme. A co-financing has been submitted
to the Agricultural Ministry and to the Ecology
Ministry.
This programme is leaded by the Consultancy Agroof

Développement and under the scientific coordination
of INRA Montpellier, with the collaboration of 16
Chambers of Agriculture and APCA and 2 others end-
users organisms.
The objective of the national

network of experimental
plots (in green) is to
complete the existing
research network (in
orange), mainly based in
the South of France. But
several provinces have
wished recently to
participate also in a near
future to this network (in
purple).
INRA-APCA Agreement
After the national end-user conference in Paris, APCA opened the discussion with INRA to
think about a future collaboration, with the signature of a possible agreement in 2005 to go
on the work realised during the SAFE programme. Face to the end-user demand, APCA
wishes a stronger participation of the Chambers of Agriculture in some Research
Development programme.

Annex 1: Letter of Luc Guyau sent to the Minister of Agriculture to defend the
eligibility of Agroforestry in the new CAP policy in France.
A l’attention de Hervé GAYMARD
Ministre de l’Agriculture, de l’alimentation, de la pêche et des affaires rurales
Paris, le 21 septembre 2004
Objet : Propositions concernant l’agroforesterie dans le cadre de l’application des accords du

Luxembourg et de la proposition de règlement européen concernant le soutien au
développement rural
Monsieur le Ministre,
L’agroforesterie connaît un écho de plus en plus favorable auprès des agriculteurs et des
collectivités territoriales. Une récente étude que nous avons menée auprès des céréaliers
des régions Centre et Poitou-Charentes montre que l’agroforesterie intéresse plus de 30 %
des personnes interrogées. De quelques projets recensés en 2001, nous devrions atteindre
la centaine de projets pour la période 2005/2006. Ainsi, pour 2005, plus de 1000 ha de
parcelles agroforestières sur terres agricoles seront mises en place en France. Ce
développement s’explique notamment par le fait que l’agroforesterie améliore la valeur
patrimoniale et les performances agro-environnementales des exploitations tout en
maintenant le revenu réel de l’agriculteur.
Le projet de règlement européen du 14 juillet 2004 prévoit de soutenir l’agroforesterie,

notamment par l’article 41, inspiré des actions de recherche développement, menés
conjointement par l’INRA et les Chambres d’Agriculture. Nous nous félicitons donc de cette
prise en compte. Nous apportons seulement quelques propositions d’amendement sur
l’écriture de certains articles, dans l’annexe ci-jointe.
Cependant, l’application des accords de Luxembourg risque d’hypothéquer le

développement de l’agroforesterie, si l’on admet au niveau français que l’emprise des arbres
est soustraite de la surface éligible aux droits.
Or, considérant :
• que l’agroforesterie est réversible et ne constitue donc pas un boisement de terres

agricoles,
• que la plupart des avantages environnementaux procurés par les parcelles
agroforestières (biodiversité, paysage, protection climatique, protection des sols et
des eaux) ne sont pas rémunérés,
• que les superficies d’emprise en jeu sont faibles au regard de la mise en œuvre de
contrôles complexes
• que la France a acquis une avance reconnue dans ce domaine au niveau européen,
Nous souhaiterions que l’activation des droits à prime soit réalisée sur la totalité de la
parcelle agroforestière et que l’emprise des arbres sur les parcelles agroforestières soit
comptabilisée dans le calcul des 3 % de surfaces enherbées (cf. annexe ci jointe).
En espérant que ces propositions retiendront toute votre attention, je vous prie d’agréer
Monsieur le Ministre, l’expression de mes sentiments respectueux
Luc Guyau, Président de l’APCA

Annexe 2 : Proposition de modification du règlement européen concernant le soutien

au développement rural (proposition du 14/07/04)
L’axe 2 du RDR au titre de l’aménagement de l’espace comporte deux sous-sections :
1. Les mesures axées sur l’utilisation durable des terres agricoles
2. Les mesures axées sur l’utilisation durable des terres sylvicoles
La sous-section 2 comprend la mesure de soutien à l’agroforesterie sur terre agricole.
Cette rédaction est ambiguë. En effet, cette mesure qui soutient l’agroforesterie sur terre
agricole se situe dans la sous-section des mesures forestières. Afin d’éviter toute confusion,
il est proposé de distinguer les deux types de systèmes agroforestiers et d’intégrer une
mesure de soutien à l’agroforesterie sur terre agricole dans la première sous-section et une
mesure de soutien à l’agroforesterie sur terre forestière dans la deuxième sous-section.
Cette distinction demande une modification des articles 34 et 41 ainsi que l’introduction
d’une nouvelle mesure donnant lieu à un nouvel article.
Modification de l’article 34
Il est ajouté un point vi à l’article 34 a). La rédaction de l’article 34 a) serait la suivante

(modifications proposées en gras) :
Article 34
L’aide prévue au titre de la présente section concerne les mesures suivantes :
a) Mesures axées sur l’utilisation durable des terres agricoles grâce à :
i) des paiements destinés aux exploitants agricoles pour les handicaps naturels en
zone de montagne;
ii) des paiements aux exploitants agricoles situés dans des zones présentant des
handicaps, autres que ceux des zones de montagne;
iii) des paiements NATURA 2000;
iv) des paiements agroenvironnementaux et en faveur du bien-être animal;
v) un soutien aux investissements non productifs.
vi) un soutien à la première installation de systèmes agroforestiers sur des terres

agricoles.
b) Mesures axées sur l’utilisation durable des terres sylvicoles grâce à :
i) un soutien au premier boisement de terres agricoles;
ii) un soutien à la première installation de systèmes agro-forestiers sur des terres

forestières;
iii) un soutien au premier boisement de terres non agricoles;
iv) des paiements NATURA 2000;

v) des paiements environnementaux forestiers;
vi) un soutien à la restauration du potentiel de production sylvicole et à l'introduction

de mesures de prévention;
vii) un soutien aux investissements non productifs.
Le point a – vi) donne lieu à nouvel article.
Proposition d’article pour l’agroforesterie sur terres agricoles
Dans la sous-section 1 (Conditions relatives aux mesures en faveur d’une utilisation durable
des terres agricoles), on ajoute un nouvel article rédigé comme suit :
Article 39
Première installation de systèmes agroforestiers sur des terres agricoles
1. Le soutien prévu à l’article 34, point a) vi), est accordée aux exploitants agricoles qui
mettent en place des systèmes agroforestiers combinant des systèmes d’agriculture
extensive et des systèmes de sylviculture.
1. L’aide couvre les coûts d'installation.
2. Par «systèmes agro-forestiers», on entend les systèmes d’utilisation des terres qui
combinent la croissance d’arbres et l’agriculture sur les mêmes terres.
3. Les sapins de Noël et les espèces à croissance rapide cultivées à court terme ne
sont pas admissibles au bénéfice de cette aide.
4. Le soutien est limité aux plafonds fixés à l'annexe I.
Modification de l’article 41 (qui devient 42)
L’article 41 de la sous-section 2 (Conditions relatives aux mesures en faveur d’une utilisation

durable des terres sylvicoles), concerne la mesure en faveur de l’agroforesterie sur terres
forestières. Il convient d’adapter le contenu actuel de l’article.
Proposition de rédaction :
Article 42
Installation de systèmes agroforestiers sur des terres sylvicoles
1. Le soutien prévu à l’article 34, point b) ii), est accordée aux exploitants agricoles qui
mettent en place des systèmes agroforestiers combinant des systèmes d’agriculture
extensive et des systèmes de sylviculture.
L’aide couvre les coûts de l’aménagement.
2. Par «systèmes agroforestiers», on entend les systèmes d’utilisation des terres qui
combinent la croissance d’arbres et l’agriculture sur les mêmes terres.
3. Les sapins de Noël et les espèces à croissance rapide cultivées à court terme ne
sont pas admissibles au bénéfice de cette aide.

4. Le soutien est limité aux plafonds fixés à l'annexe I.
Annexe 3 : Propositions concernant l’agroforesterie dans le cadre de l’application des

accords du Luxembourg
Etat actuel
Le document de travail AGRI/2254/2003 recommande que le seuil pris en compte pour

caractériser une parcelle arborée soit de 50 tiges par ha. Au-delà, la parcelle devient
inéligible au titre du PU sauf dérogation pour des motifs agro-environnementaux.
Il est également spécifié dans le règlement 1782/03 que l’agriculteur perd ses droits à
paiements pour les surfaces mises en cultures pérennes (article 51).
Néanmoins le premier principe de l’article 8 du règlement d’application 796/2004 spécifie qu’

« une parcelle boisée est considérée comme une parcelle agricole aux fins du régime d’aide
« surfaces » sous réserve que les activités agricoles visées à l’article 51 du règlement (CE)
n° 1782/2003 ou, le cas échéant, que la production envisagée puissent se dérouler comme
elles se dérouleraient sur des parcelles non boisées situées dans la même zone. »
Proposition
Compte-tenu que :
L’agroforesterie répond aux 4 objectifs fixés par les conditions de bonne pratique agricole et
environnementale, à savoir :
• Protection contre l’érosion des sols grâce au maillage des lignes d’arbres enherbées,
• Maintien de la matière organique sous le double effet de l’enherbement et de la

décomposition du feuillage et des racines annuelles
• Maintien de la structure des sols
• Niveau minimum d’entretien, assuré par les animaux dans les zones sylvopastorales.
Et que d’autre part, l’agroforesterie répond aux enjeux définis par les directives européennes
sur l’environnement, en particulier les directives concernant la préservation de la qualité de
l’eau (directive 91/676) et a la directive sur le bien-être des animaux (directive 98/58)
Il est proposé que :
Nous souhaiterions que l’activation des droits à prime soit réalisée sur la totalité de la
parcelle agroforestière.
Pour cela, la parcelle agroforestière devra respecter les normes usuelles en agroforesterie, à
savoir que la parcelle doit être majoritairement agricole (culture ou pâture) et que la densité
d’arbres soit comprise entre 50 et 200 arbres par ha.
Les arbres double-fin, cultivés pour le bois et pour leur production fruitière, sont éligibles à
condition que la hauteur de bille soit supérieure à 2 m et nette de tout point de greffage sur
cette hauteur. Conformément à la réglementation, l’exploitant ne pourra cumuler différentes
aides sur cette surface. Soit il opte pour la déclaration de la surface dans le cadre du RPU,
soit il opte pour une déclaration de surface en verger. Dans ce dernier cas, la parcelle n’est

plus éligible aux droits à prime mais peut prétendre aux aides vergers (ex aides aux fruitiers
à coque).
Remarque :
L’introduction d’arbres à faible densité ne modifie pas le montant global des primes reçues
par l’exploitation. Dans cette démarche, il n’y a aucune possibilité de transgresser les textes
d’application afin de percevoir davantage de primes : un projet agroforestier n’augmente ni
ne diminue le niveau global de paiements perçus par l’exploitation.
Eco-conditionnalité
Parmi les dispositions que la France a prises au titre de la conditionnalité des aides, figure
l’obligation d’implanter des bandes enherbées ou un couvert d’intérêt environnemental sur
une surface équivalente à 3 % de leur surface SCOP. La largeur des bandes doit être
comprise entre 5 et 10 m.
Proposition
Il est proposé que les lignes d’arbres enherbées plantées dans les parcelles agroforestières
puissent être comptabilisées dans le calcul des 3 %. Dans ce cas, une dérogation est
demandée afin que la largeur des lignes d’arbres en agroforesterie puisse se situer sous le
seuil des 5 mètres demandés dans le cas des bandes enherbées.
Annexe 3 : Normes agroforestières françaises
Une parcelle agroforestière doit techniquement satisfaire les deux critères suivants :
1. Entre 50 et 200 arbres/ha répartis sur l’ensemble de la parcelle
2. Plus de 50 % de la surface de la parcelle cultivée. Cette culture intercalaire

doit pouvoir être effectuée dans des conditions comparables à celles des
parcelles non arborées de la région.

Annex 4: APCA proposal for the consideration of agroforestry in the CAP

See document word: “APCA proposal CAP.doc”
Annex 5: Land Status for agroforestry

See document word: “French agroforestry status by APCA.doc”

University of Thessaloniki Report
15 Contractor 10: University of Thessaloniki

Contractor 10 : University of Thessaloniki (GREECE)

Assemblée Permanente des Chambres d’Agriculture, 9, Avenue Georges V - 75008 PARIS, France
Department Politiques Territoriales et Stratégie Environnementale
(Land Management and Environmental Strategies Branch)
Scientific team and time spent of the WPs

Name Tel Fax E-mail

Vasilios Papanastasis
Workpackage 2: European silvoarable knowledge
• Start date: Month 37 (01 August 2004)
• Completion date: Month 42 (31 January 2005)
• Current status: Active
• Partner responsible: Vasilios Papanastasis, AUTH
• Person months of WP2: 0.5
INRA WU NERC LEEDS CRAN CNR UEX FAL APCA AUTH Total
Technical Annex 14
Month 1-6 2
Month 7-12 5
Month 13-18 1
Month 18-24 0.5
Month 25-30 2.5
Month 31-36 2.5
Month 36-42 0.5
Total 42 Months 14
Objectives
Objectives
AUTH will study Greek traditional silvoarable systems in order to collect technical and
economic data about the productivity and viability of these systems. They will include
structural data about intercropping systems in the area.
The farmers’ awareness and adaptability of agroforestry.
Work carried out in the reporting period
Completion of survey on still living silvoarable systems

The survey on still living silvoarable systems at selected sites around Greece was completed
during the reporting period. Corrections were made in the database of these systems and
additional information such as photos and maps for the surveyed systems were provided.
The Greek team also contributed to the preparation of the paper ‘Silvoarable agriculture in
Europe – past, present and future’ that describes the still living silvoarable systems around
Europe.
A review paper for agroforestry systems in Greece prepared by the Greek team was
presented to the 4th Pan-Hellenic Rangeland Congress held in November at Volos,
Greece. This presentation was based on the survey of still living agroforestry
systems at the Askio Municipality and on the preview work related to these systems.
The paper is in Greek with English summary which is attached. The paper is going to
be included in the proceedings of the Congress.
Future Actions
• After the end of the project a proposal for the detailed survey of agroforestry
systems in Greece will be submitted for funding to the Ministry of Rural
Development and Food (the new name for the Ministry of Agriculture).
AGROSILVOPASTORAL SYSTEMS IN GREECE
K. Mantzanas, E. Tsatsiadis, I. Ispikoudis and V.P. Papanastasis
Laboratory of Rangeland Ecology, Aristotle University, P.O. Box 286,
54124 Thessaloniki, Greece. e-mail: konman@for.auth.gr
Summary
Agrosilvopastoral systems traditional and new occur in several parts of Greece and play an
important role in local economy. They also play a very important ecological role because they
prevent soil erosion and surface runoff while they improve the landscape and conserve the
biodiversity. They separate to several types according to tree species. Over the last decades,
these systems have been substantially reduced due to several reasons including both
extensification and intensification. A measure for the preservation of these systems could be
a national survey. For this reason the area of Askio Municipality was selected as a study
case. Several combinations of trees and crops were identified in that survey conducted
during the last three years. The main tree species were oaks, walnuts, poplars, and fruit
trees. The understorey crops consisted mainly of cereals (wheat, barley, corn), lucerne,
tobacco and vegetables. Crops are used for livestock feeding directly (grain, hay) or
indirectly (grazing). In the latter case, animals use these areas in the critical periods of the
year such as the summer and early autumn (after the crop harvest).
Key words: Trees, crops, grazing, reasons for reduction, survey
Workpackage 9: Developing European guidelines for policy implementation
Start date: Month 37 (01 August 2004)
Completion date: Month 42 (31 January 2005)
Current status: Active
Partner responsible: Vasilios Papanastasis, AUTH

Person months of WP9: 2
INRA WU NERC LEEDS CRAN CNR UEX FAL APCA THES Total
Technical
2
Annex
Month 1-6
Month 7-12
Month 13-18 1
Month 18-24 1.5**
Month 25-30 1
Month 31-36 2*
Month 37-42 2*
Total 42 Months -5.5
* from wp8, ** from wp1
1.1. Objectives
WP9 will produce a synthesis report on Silvoarable Agroforestry in the context of economic
and social changes to agricultural and forestry policies being implemented in Agenda 2000
(e.g. 1257/99), and provide guidelines to Member States and Autonomous Regions on the
potential uptake of agroforestry systems. It will describe the effect of subsidiarity on farm-
forestry practices, and use the socio-economic model to investigate the potential effects on
agroforestry uptake of different legal structures, financial incentives and market price levels.
CURRENT TASKS
Local, regional, national and European policies will be reviewed in relation to agroforestry.
The information collected will be analysed and suggestions will be formulated in
collaboration with other partners. The final document on agroforestry policy will be
reviewed to address the issues of Mediterranean countries, and to promote the use of
traditional silvoarable systems if these systems have proved to be of a productive or
environmental value.
1.2. WORK CARRIED OUT IN THE REPORTING PERIOD
1.2.1. Policy aspects
After reviewing the EU regulations related to agricultural policy and the environment
considerable time was spent to meet people and discuss the promotion of silvoarable
agroforestry and agroforestry in general. Since agroforestry is not included explicitly
in any of the relevant regulations, several Greek authorities were contacted to
discuss how this land use practice can be accommodated so that traditional
silvoarable systems are maintained and preserved and new ones are established.

The question that arose was whether silvoarable practices should be considered as
part of the good agricultural and environmental practices.
The SAFE project was presented at the 4th Panhellenic Rangeland Congress held in
November of 2004 at Volos, Greece. The various scientists attending the congress
followed with real interest and enthusiasm the presentation for the agroforestry
systems and the work done at the Municipality of Askio.
The preparation of the end-users national conference started last August. The idea
was to hold it in the framework of the international fair ZOOTECHNIA in order to be
attended by several officers, politicians, scientists and farmers. Contacts were made
with the various Services of the Ministry of Rural Development and Food, involved
with agricultural and forestry policy. Invitations for attending the conference were sent
to various Organisations and Institutes around Greece such as the Municipalities of
the Kozani prefecture, Forest offices, Research Institutes, non governmental
organisations, the Ministry of Rural Development and Food in Athens, Universities
and Technological Institutes as well as farmer unions and individual farmers. A press
release was sent to all major newspapers, TV channels and radios in order to
advertise the conference. Additionally, Mr A. Christidis, officer of the European
Commission was invited to give a presentation about the new European Rural
Development Regulation and the involvement of agroforestry systems (Article 41).
The other speakers were Prof. V. Papanastasis, who introduced the silvoarable
systems and the SAFE project; the coordinator of the project (Dr C. Dupraz), who
presented the results of the project; As. Prof. I. Ispikoudis, who analysed the
historical and cultural perspectives of silvoarable systems in Greece; Dr K.
Mantzanas, E. Tsatsiadis and E. Mpatianis, who presented the extant silvoarable
systems in Greece and the work done in Askio; and finally Mrs A. Logothetou from
the Ministry of Rural Development and Food, who presented the agri-environment
measures and codes of good agricultural and environmental practices. A dossier was
prepared with the pamphlet of traditional and new silvoarable systems of Askio, the
translation in Greek of the info paper at the European Research Magazine and a
proposal for forming a Greek agroforestry forum, which was given to all the
participants. Overall, 150 participants attended the workshop and a lengthy
discussion was held concerning the future of silvoarable systems in Greece.
1.2.2. Community agroforestry plots
The exceptionally dry summer period resulted in a number of dead trees in the three
experimental plots established last year. Specifically, the first plot (Siargas’)
intercropped with maize did not have any dead trees due to drought because it was
irrigated during the summer period (July-October 2003 and 2004). It had only 2 dead
trees out of the 43 planted destroyed by the harvest machine (Figures 1 and 2). The
second plot (Tsatsiadis’) intercropped with wheat had 3 dead trees and 7 trees with
dead leaves out of the 28 planted (Figures 5 and 6). They were replaced in
November of 2004 with trees from the Forest Service nursery of Thessaloniki. Finally,
the third plot (Strebas’) intercropped with wheat had 27 dead trees and 12 trees with
dead leaves out of the 63 planted (Figures 3 and 4). They were also replaced in
November of 2004. Detailed information for the experimental plots is given in tables
1-3.
Table 1. Siargas’ experimental plot

Owner: Antonios Siargas

Farmer: Antonios Siargas
Crop Maize
Ploughing Date November 10, 2004
Depth
Seeding Date
Quantity
Fertilization Basic
Surface
Herbicide Date
application Herbicide
Quantity
Irrigation Dripping
Yield
Table 2. Tsatsiadis’ experimental plot

Owner: Antonios Tsatsiadis
Farmer: Aristides Tremmas
Crop Barley
Ploughing Date November 2004
Depth 0.25-0.30 m
Seeding Date November 2004
Quantity 300 kg/ha
Fertilization Basic November 2004, type 20-16-0,
quantity: 300 kg/ha
Surface
Herbicide Date With seeding
application Herbicide Dablitine
Quantity 1 kg per tone of seeds
Irrigation No
Yield
Table 3. Strebas’ experimental plot

Owner: Ioannis Strebas
Farmer: Athanasios Zois
Crop Barley
Ploughing Date November 2004
Depth 0.25-0.30 m
Seeding Date November 2004
Quantity 300 kg/ha
Fertilization Basic November 2004, type 20-16-0,
quantity: 300 kg/ha
Surface
Herbicide Date With seeding
application Herbicide Dablitine
Quantity 1 kg per tone of seeds
Irrigation No
Yield

Figure 1. Siargas’ experimental plot (August 2004)
Figure 2. Siargas’ experimental plot (January 2004)

Figure 3. Strebas’ experimental plot (August 2004)
Figure 4. Strebas’ experimental plot (January 2004)

Figure 5. Tsatsiadis’ experimental plot (August 2004)
Figure 6. Tsatsiadis’ experimental plot (January 2004)

NATIONAL WORKSHOP
SILVOARABLE SYSTEMS IN GREECE: TECHNICAL AND POLICY

CONSIDERATIONS
Friday, 4 February 2005
HELLEXPO – Thessaloniki-Greece
PROGRAMME
9:00 Registration
9:45 Addresses
Session I: Results of the SAFE project
10:00 Introduction to silvoarable systems: the SAFE project
V. Papanastasis, Aristotle University of Thessaloniki
10:30 Modern silvoarable systems in Europe and their potential uptake: lessons from the
SAFE project
C. Dupraz; SAFE Coordinator, France
11:15 Extant silvoarable systems in Greece: the case of Municipality of Askio
K. Mantzanas, E. Tsatsiadis and E. Mpatianis, Aristotle University and Municipality of

Askio
12:00 Break
Session II: Silvoarable systems and agricultural policy
12:30 Silvoarable systems of Greece: an historical and culture perspective
I. Ispikoudis, Aristotle University of Thessaloniki
13:15 Agri-environment measures and codes of good agricultural practice
A. Logothetou, Ministry of Rural Development and Food
14:00 Agroforestry and the new CAP in the European Union
A.Christidis, European Commission
14:45 Discussion and conclusions
16:00 End of the workshop

Contribution of the sub-contractor to AUTH: Municipality of ASKIO
Department involved: Bureau of Planning and Development
Implementation team
Name Tel. Fax E-mail address
E. Batianis +30 24650 71062 +30 24650 71329 askio@otenet.gr
During the reporting period (1st of August 2004 – 31st of January 2005) of the project,
we carried out the following activities related to the SAFE project:
We helped the AUTH team to collect the additional information needed for the
productivity and economic efficiency of the traditional systems in our region.
We informed all the Forest and Agricultural services, the Municipalities and the
Prefectures of Region of Western Macedonia about the SAFE project and we
dispatched to them a document with regard to the program with the pamphlet it has
been published by AUTH.
We participated in the national workshop, which was held in Thessaloniki on 4

February 2005.
Kaloneri, February 2005
Vasilios Patras
Mayor

SAFE Fourth Year Annual Report Volume 3

Uploaded by

Copyright:

Available Formats

You might also like

SAFE Fourth Year Annual Report Volume 3

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SAFE Fourth Year Annual Report Volume 3

Uploaded by

Copyright:

Available Formats

Quality of Life and Management of Living Resources

Silvoarable Agroforestry For Europe (SAFE)

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 4

Relative crop yield

Arable 156 trees/ha

SAFE PROJECT FINAL PROGRESS REPORT

SAFE Final Progress Report – Volume 3 – May 2005 3

SAFE Final Progress Report – Volume 3 – May 2005 4

SAFE Final Progress Report – Volume 3 – May 2005 5

1 Contractor 1: INRA – SYSTEM

Contractor 1: INRA – SYSTEM (FRANCE)

Scientific team and time spent on the WPs

Name. Unit Tel Fax E-mail

SAFE Final Progress Report – Volume 3 – May 2005

SAFE Final Progress Report – Volume 3 – May 2005 7

1. silvoarable agroforestry, comprising widely-spaced trees and/or shrubs associated

2. silvopastoral agroforestry, a combination of trees, forage (pasture) and livestock

3. forest gardening, comprising multi-species and multi-storied dense plant associations,

4. forest farming, the cultivation of edible, medicinal or decorative specialty crops as

SAFE Final Progress Report – Volume 3 – May 2005 8

WP3 Silvoarable experimental network

SAFE Final Progress Report – Volume 3 – May 2005 9

Site Tree Year of Crop in 2003- Compared AF treatments

Climate and environmental site characterisation at both INRA sites

Climate and microclimate recording

Soil humidity and water table

Tree size and phenological development

SAFE Final Progress Report – Volume 3 – May 2005 10

The crop measurements

SAFE Final Progress Report – Volume 3 – May 2005 11

Table 7: Calendar of crop measurements in agroforestry plots at Vézénobres and

SAFE Final Progress Report – Volume 3 – May 2005 12

Tree canopy pru ned Tree m icro-p lot

Table 8: Field layout with measurement points (micro-plots) at Vézénobres

Material and Method

SAFE Final Progress Report – Volume 3 – May 2005 13

Climate and microclimate characterisation

SAFE Final Progress Report – Volume 3 – May 2005 14

Durum wheat management

2 tree row orientations * 2 plot orientation * 2 pruned/ unpruned * 2 distances

All measurements were done on all treatments (micro-plots).

SAFE Final Progress Report – Volume 3 – May 2005 15

Climate and environmental site characterisation

Daily m ax. temperature

SAFE Final Progress Report – Volume 3 – May 2005 16

Day of year from start 2003 to end 2004

Table 13: Daily global radiation at Vézénobres in season 2003- 2004

Agroforestry treatment Yield in agroforestry (AF) Yield in Ratio

SAFE Final Progress Report – Volume 3 – May 2005 17

SOUTH NORTH WEST EAST

SAFE Final Progress Report – Volume 3 – May 2005 18

Summary and conclusion

The major conclusions that can be drawn are:

SAFE Final Progress Report – Volume 3 – May 2005 19

SAFE Final Progress Report – Volume 3 – May 2005 20

WP4 Above-ground interactions

SAFE Final Progress Report – Volume 3 – May 2005 21

WP5 Below-ground interactions