guiding douglas-fir seed selection in europe under ...m.… · spain france & belgium central...

Guiding Douglas-fir Seed Selection in Europe Under Changing Climates

Miriam Isaac-Renton1, David Roberts1, Andreas Hamann1, Heinrich Spiecker2

Funding Acknowledgements: Natural Sciences and Engineering Research Council of Canada, the Commission of the European Union, Human Resources and Skills De-velopment Canada, and the Universities of Alberta and Freiburg.

1) Dept. of Renewable Resources, University of Alberta, Canada2) Chair of Forest Growth, University of Freiburg, Germany* isaacren@ualberta.ca

IntroductionDouglas-fir is an important tree species in its native range in North America, and in Europe,

where it is was introduced more than 185 years ago for its high productivity and wood qual-

ity. Many early plantations have known seed source origins, which today allow them to

serve as a unique experimental test bed to investigate how trees react to being transferred

to new climate conditions.

MethodsWe evaluate more than 700 Douglas-fir provenances from 13 North American re-

gions at 120 European test sites (Fig 1). Model predictions of where the 13 North

American regions find their best climate match in Europe under current and future

climate were carried out with the RandomForest ensemble classifier, a widely

used bioclimate envelope modeling approach.

76

9

8

5

21

3

4

97 98 96

9192

9593

86-9094

62

3231

99

10

11 30

35-4041-47

48-59

70-75

33

34

19-21

2922-28

6967-68

76

8385

78

77

81

8079

8482

18

6061

12

13171615

14

6563

64 66

120117

118119

108107

109110

112

115113

114116

101

100

106102105

111103

FinlandNorwayScotlandS. UKN. Coast

PolandE. GermanySpainFrance & BelgiumCentral GermanyS. Germany

ItalyCentral EuropeCoastal BalkansBalkansRomaniaTurkey

Planting Sites

Spain300 20

10

0

-10

200

100

0Prec

ipita

tion

(mm

)J F M A M J J A S O N D

Southern UK300 20

10

0

-10

200

100

0Prec

ipita

tion

(mm

)

J F M A M J J A S O N D

Scotland300 20

10

0

-10

200

100

0Prec

ipita

tion

(mm

)

J F M A M J J A S O N D

Poland300 20

10

0

-10

200

100

0J F M A M J J A S O N D

Tem

pera

ture

(°C

)

North Coast300 20

10

0

-10

200

100

0J F M A M J J A S O N D

France & Belgium300 20

10

0

-10

200

100

0J F M A M J J A S O N D

Finland300 20

10

0

-10

200

100

0

Tem

pera

ture

(°C

)

J F M A M J J A S O N D

Central Germany300 20

10

0

-10

200

100

0Prec

ipita

tion

(mm

)

J F M A M J J A S O N D

Central Europe300 20

10

0

-10

200

100

0J F M A M J J A S O N D

Balkans300 20

10

0

-10

200

100

0J F M A M J J A S O N D

Tem

pera

ture

(°C

)

Results & ApplicationsModel predictions of optimal climate matches for western and central Europe con-

formed well with observed provenance performance (Fig 2, first 10 rows, Finland

to Italy). Under eastern Europe’s continental climates, however, the results were

mixed (Fig 2, rows 11-17), and a safe strategy would be to use slightly less pro-

ductive provenances with reasonable climate matches (provenances that are both

above average in Fig 2a and toward the right in Fig 2b).

Given adequate model performance at least for western Europe, model projec-

tions for observed and projected climate change may provide guidance for Doug-

las-fir plantation forestry (Fig 3). Both recent climate change trends and projec-

tions for the 2020s suggest that European foresters should switch to provenances

from more southern or drier origins in North America.

0.0001 0.001 0.1 1Random forest class probability

TurkeyRomaniaBalkans

Coastal BalkansCentral Europe

PolandEastern Germany

ItalySpain

France & BelgiumSouthern Germany

Central GermanyNorth Coast

Southern UKScotlandNorwayFinland

TurkeyRomaniaBalkans

Coastal BalkansCentral Europe

PolandEastern Germany

ItalySpain

France & BelgiumSouthern Germany

Central GermanyNorth Coast

Southern UKScotlandNorwayFinland 4

14814259101332272854

848358120017718891217258276911571031

0.460.520.430.310.270.170.350.300.200.330.480.500.370.680.320.640.60

Nsites Nprov SE

0.471.160.410.76-0.740.420.801.230.530.780.86-1.67-0.740.04-1.43-1.951.46

1st

-1.150.760.630.441.561.180.72-0.270.681.260.380.760.790.911.160.780.45

2nd

0.670.340.540.91-0.840.751.120.95-0.26-0.34-0.851.460.40-0.12-0.710.43-0.78

3rd

−2 0 1 2−1Standard deviation from mean height

Observed

Predicted

BC CoastWA CoastOR CoastCA LowElevationWA Dry CoastOR Dry CoastWA Coast CascadesOR Coast CascadesCA High ElevationInterior CascadesInterior NorthInteriorInterior South

Provenances

Fig 1. Map of Europe with numbered planting sites coloured by region, with cli-mate diagrams displaying average monthly temperature (°C) and average

Fig 2.The upper dot plot (a) displays the observed perfor-mance of North American populations in European re-gions. Performance is reported in units of standard deviation from the site mean. Statistics on the right indicate the number of planting the number of provenances and the aver-age standard error for dot plot entries in each row. The lower dot plot (b) displays the climate match (better match on right) of the North American populations in the European regions. The columns on the right indicate the productivity gain or loss if the first, second, or third model recommendation would be planted based on provenance trial performance.

A2 2080

A2 2050A2 2020

1995-2009Observed

1961-1990Observed

ProvenancesBC CoastWA CoastOR CoastCA Low ElevationWA Dry CoastOR Dry Coast

WA Coast CascadesOR Coast CascadesCA High ElevationInterior CascadesInterior NorthInteriorInterior South

ProvenancesBC CoastWA CoastOR CoastCA Low ElevationWA Dry CoastOR Dry Coast

WA Coast CascadesOR Coast CascadesCA High ElevationInterior CascadesInterior NorthInteriorInterior South

Fig 3. Predictions of climatically best matching North American Douglas-fir populations as projected using a biocli-matic envelope model for past climate (1961-1990), a recent 15-year climate average (1995-2009) and the 2020’s, 2050’s and 2080's under an A2 scenario.

ObjectiveHere, we test if models developed to guide assisted migra-

tion of Douglas-fir in North America under climate change

can accurately predict the success of provenance transfers

to Europe. We then use a validated model to guide seed

transfer under climate change in Europe.