1

Appendix S1


Method of how modern forest maps were used to assign a modern equivalent of composition to each historical observation. These 1:20 000 maps delimit photo-interpreted stand polygons by identifying the main taxa present in the canopy (Berger 2008). These stand polygons were used to construct both dominance and presence maps of the nine taxa (presence maps represent areas where a taxon is present, regardless of whether it is dominant or just present). The following map also illustrates the j polygonsas intersections between 100-meter buffers and Thiessen polygons (Brassel and Reif 1979) that weregenerated by the historical observations points or line centers.

Taxa rank of abundance are assigned on the basis of their relative coverage in term of dominance and presence within the j polygons:

  • Rank 4 is assigned to the taxon with the maximum relative coverage of dominance in j.
  • Rank 3 is assigned to the second taxon with the maximum relative coverage of dominance, only if this taxa dominance covers ≥ 30 % of the forested area; otherwise rank 3 is assigned to the taxon with the maximum relative coverage of presence in j.
  • Rank 2 is assigned to the taxon with the maximum relative coverage of presence in j if it is not already ranked as 3,otherwise rank 2is assigned to the second taxon with the maximum relative coverage of presence in j.
  • Rank 1 is then assigned to all taxa covering ≥ 10% in terms of dominance or ≥ 25 % in terms of the presence of the forested area in j.
  • Rank 0 is assigned to all taxa covering < 10% in terms of dominance or < 25 % in terms of presence of the forested area in j.

References

Berger J.P. (2008) Norme de stratification écoforestière. Quatrième inventaire écoforestier. Ministère des Ressources naturelles et de la Faune du Québec, Québec, Canada.

BrasselK.E. and Reif D. (1979) A Procedure to Generate Thiessen Polygons.Geographical Analysis, 11, 289–303.

Appendix S2

Maps of environmental variables that were derived from modern forest maps (Berger 2008).

Appendix S3

Spatial structure of β-diversity assessed by number of significant dbMEM after forward selection (histogram) and adjusted R2(curves) for successive blocks of 20 dbMEM, representing the gradation from the broadest (block 1) to the finest scales (block 10).


Here we used the dbMEMs to assess on what scale the forest composition was spatially structured. During the production of dbMEMs with the “PCNM” function, dbMEMs are automatically ordered by decreasing spatial scale (Legendre et al. 2009; 2013). Thus, we decomposed the spatially structured β-diversity (i.e. pure spatial fraction plus the shared space-environment fraction) for successive blocks of 20 dbMEMs, which allows to determine whether β-diversity is rather related to broad-scaled or to finer-scaled spatial structure. Results clearly show that both pre-industrial and modern communities tend to form large landscape units of homogeneous composition at the regional scale than finer-scaled patches at the local scale.

References

Legendre P., Borcard D., Blanchet F.G., and Dray S. (2013) PCNM: MEM spatial eigenfunction and principal coordinate analyses. R package version 2.1-2/r109.

Legendre P., Mi X., Ren H., Ma K., Yu M., Sun I.-F., and He F. (2009) Partitioning beta diversity in a subtropical broad-leaved forest of China. Ecology, 90, 663–674.