ElectronicSupplementary Material for Paal et al.Both spatiotemporal connectivity and habitat quality limit the immigration of forest herbs into wooded corridors

Fig. S1 Relationship between present day and historical forest cover (from ca 1890) in a 250 m radius around forest edge and corridor sampling points. r – Pearson’s correlation coefficient, p – p value

Table S1 Species list of forest specialists and forest generalists used in the study. F – forest specialist; G – forest generalist

Species / Group / Observed in forests (%) / Observed in corridors (%)
Actaea spicata / F / 30.2 / 12.0
Aegopodiumpodagraria / G / 59.4 / 83.1
Anemone nemorosa / F / 55.2 / 19.3
Angelica sylvestris / G / 29.2 / 12.0
Asarum europaeum / F / 25.0 / 3.6
Athyriumfilix-femina / F / 65.6 / 8.4
Calamagrostisarundinacea / F / 19.8 / 3.6
Carexcespitosa / G / 10.4 / 4.8
Carexdigitata / F / 25.0 / 1.2
Carexelongata / G / 11.5 / -
Carexpallescens / G / 11.5 / 2.4
Carexsylvatica / F / 21.9 / 1.2
Circaeaalpina / F / 11.5 / -
Cirsiumoleraceum / G / 28.1 / 8.4
Convallaria majalis / F / 22.9 / 19.3
Crepispaludosa / G / 49.0 / 4.8
Deschampsiacespitosa / G / 42.7 / 45.8
Dryopteriscarthusiana / F / 81.3 / 7.2
Dryopterisfilix-mas / F / 55.2 / 10.8
Roegneriacanina / F / 14.6 / 4.8
Epilobiumangustifolium / G / 13.5 / 13.3
Epilobiummontanum / G / 22.9 / 31.3
Equisetum pratense / G / 42.7 / 38.6
Equisetum sylvaticum / F / 18.8 / 6.0
Festucagigantea / F / 13.5 / 13.3
Filipendulaulmaria / G / 41.7 / 26.5
Fragariavesca / G / 63.5 / 57.8
Galeobdolon luteum / F / 34.4 / 6.0
Galiumpalustre / G / 20.8 / 6.0
Geranium sylvaticum / G / 14.6 / 10.8
Geum rivale / G / 54.2 / 32.5
Geum urbanum / G / 51.0 / 65.1
Gymnocarpiumdryopteris / F / 24.0 / -
Hepatica nobilis / F / 24.0 / 3.6
Impatiens noli-tangere / F / 13.5 / 1.2
Impatiens parviflora / G / 59.4 / 50.6
Juncuseffusus / G / 12.5 / 1.2
Luzulapilosa / F / 47.9 / 6.0
Lysimachia vulgaris / G / 36.5 / 26.5
Maianthemumbifolium / F / 42.7 / 10.8
Melampyrumnemorosum / G / 12.5 / 14.5
Melicanutans / F / 22.9 / 3.6
Mercurialisperennis / F / 12.5 / 2.4
Moehringiatrinervia / F / 19.8 / 20.5
Mycelismuralis / G / 41.7 / 9.6
Oxalis acetosella / F / 82.3 / 10.8
Paris quadrifolia / F / 78.1 / 22.9
Poanemoralis / F / 13.5 / 37.3
Poatrivialis / G / 17.7 / 36.1
Ranunculuscassubicus / G / 32.3 / 30.1
Ranunculusrepens / G / 28.1 / 25.3
Rubus idaeus / G / 77.1 / 32.5
Rubus saxatilis / F / 49.0 / 22.9
Scrophularia nodosa / G / 15.6 / 18.1
Scutellariagalericulata / G / 11.5 / 4.8
Solidago virgaurea / G / 28.1 / 27.7
Stachys sylvatica / F / 16.7 / 1.2
Stellariaholostea / F / 17.7 / 6.0
Stellarianemorum / F / 15.6 / -
Taraxacum officinale / G / 25.0 / 92.8
Trientaliseuropaea / F / 29.2 / -
Urtica dioica / G / 43.8 / 59.0
Vacciniummyrtillus / F / 18.8 / -
Valeriana officinalis / G / 18.8 / 22.9
Veronica chamaedrys / G / 39.6 / 91.6
Veronica officinalis / G / 12.5 / 6.0
Viola mirabilis / F / 24.0 / 3.6
Viola palustris / G / 14.6 / 1.2
Viola riviniana / F / 33.3 / 8.4

Fig. S2 Spearman correlation coefficients between variables that were used in model building

Fig. S3 PCA ordination biplot of habitat structural and landscape connectivity characteristics. Vectors illustrate the correlations with the first and second principal component. Tree species are presented in italics, overstorey trees are distinguished from understorey trees by the four first letters of genus name being in capitals. Variables that are included in the models are presented in bold, superscript letters denote the species group model where the variable is used. S – forest specialists, G – forest generalists

Fig. S4 PCA ordination biplot for site scores

Factor analysis

Prior to model building, we conducted a Factor analysis (function fa of the psych package; Revelle 2014)on the set of variables that were considered for predictors. We used varimax rotation and the principal factor method to construct the factors. The optimal solution comprised of two factors. The factor analysis result was taken into account in model building, i.e. variables that had a loading of > 0.5 and loaded into the same factor were not included into the models simultaneously.

Table S2 Results of factor analysis of studied landscape and habitat variables. The first factor explained 12% and the second 9% of the total variance.Absolute loadings > 0.5 are marked as bold.Superscript letters denote the species group model where the variable is used. S – forest specialists, G – forest generalists

Long name / Short name in Figures / Factor 1 / Factor 2
Landscape structure:
Forest area in 250m bufferG / %forest.b250mG / 0.56 / 0.42
Residential area in 100m buffer / %hum.area.b100m / -0.52 / -0.14
Agricultural area in 250m buffer / %agricult.area.b250m / 0.04 / -0.39
Length of ditches in 250m buffer / len.ditches.b250m / 0.10 / 0.50
Length of paved roads in 250m buffer / len.paved.road.b250m / -0.57 / 0.00
Length of unpaved roads in 250m buffer / len.unpav.road.b250m / -0.40 / -0.14
Habitat structure:
Canopy coverS,G / Canopy coverS,G / 0.46 / -0.11
Understorey tree cover / Understorey tree cover / 0.46 / 0.03
Habitat widthS,G / Habitat widthS,G / -0.13 / -0.02
Road / Road / -0.74 / -0.24
Paved road / Paved road / -0.72 / -0.24
Overhanging branchesS / Overhanging branchesS / 0.05 / -0.07
Ditch2 / Ditch2 / 0.14 / 0.46
Disturbance / Disturbance / 0.23 / -0.17
Maximum tree diameter / Max. tree diameter / -0.65 / -0.29
Historical forest area in 250mbufferS / %hist.forest.b250mS / 0.32 / 0.35
Historical woodland area in 250m buffer / %hist.woodland.b250m / 0.38 / 0.38
Average tree heightS / Aver. tree heightS / -0.07 / 0.15
CV tree height / CV tree height / 0.07 / 0.33
Edge length / Edge length / -0.65 / -0.21
Soil:
Soil texture: sand / Soil texture: sand / -0.03 / -0.11
Soil type: moist / Soil type: moist / 0.09 / 0.79
Cambisol / Cambisol / -0.02 / -0.12
Gleysol / Gleysol / 0.07 / 0.85
Luvisol / Luvisol / 0.02 / -0.71
Podzol / Podzol / -0.10 / 0.05
Tree layer:
Populustremula / POPUtrem / 0.25 / 0.17
Betula pendula / BETUpend / 0.12 / 0.37
Picea abies / PICEabie / 0.30 / 0.17
Alnusincana / ALNUinca / 0.55 / 0.21
Tiliacordata / TILIcord / -0.53 / -0.13
Salix spp / SALIsp / 0.23 / -0.04
Fraxinus excelsior / FRAXexce / -0.14 / -0.16
Quecusrobur / QUERrobu / -0.40 / -0.16
Padus avium / PADUaviu / 0.22 / -0.06
Acer plantanoides / ACERplat / -0.19 / -0.32
Shrub layer:
Populustremula / poputrem / 0.04 / 0.24
Betula pendula / betupend / 0.04 / 0.26
Picea abies / piceabie / 0.19 / 0.19
Alnusincana / alnuinca / 0.56 / 0.15
Tiliacordata / tilicord / -0.19 / -0.02
Salix spp / salisp / 0.25 / 0.23
Fraxinus excelsior / fraxexce / 0.04 / -0.23
Quecusrobur / querrobu / -0.07 / 0.05
Padus avium / paduaviu / 0.60 / -0.09
Acer plantanoides / acerplat / -0.03 / -0.22
Sorbusaucuparia / sorbaucu / 0.31 / -0.10
Corylusavellana / coryavel / 0.01 / -0.12

Revelle W (2014) psych: Procedures for Personality and Psychological Research. Northwestern University, Evanston, Illinois, USA

Table S3 Results of the linear regression (LM) model of habitat structure and landscape factors on the species richness of forest specialists and forest generalists. The general linear model with transformed species richness (to optimize the normality of residuals) was built to calculate the metric of relative importance of drivers as partial eta-squared (ηp2) for each variable in the model, and is presented here as comparison to the main GLM model. df – degrees of freedom, F – F statistic, p – p value.

ηp2 / df / F / p
Forest specialists
Intercept / 0.011 / 1 / 2.06 / 0.153
Habitat connectivity type / 0.153 / 2 / 16.47 / <0.001
Ancient woodland area (r=250m) / 0.064 / 1 / 12.43 / <0.001
Habitat width / 0.059 / 1 / 11.49 / <0.001
Canopy cover / 0.037 / 1 / 6.96 / 0.009
Canopy cover2 / 0.031 / 1 / 5.77 / 0.017
Aver. canopy height / 0.062 / 1 / 11.84 / <0.001
Overhanging branches / 0.069 / 1 / 12.52 / <0.001
Error / 164
Forest generalists
Intercept / 0.028 / 1 / 4.63 / 0.033
Habitat connectivity type / 0.021 / 2 / 1.72 / 0.182
Present-day forest area (r=250m) / 0.069 / 1 / 12.05 / <0.001
Habitat width / 0.032 / 1 / 5.36 / 0.022
Canopy cover / 0.024 / 1 / 4.04 / 0.046
Canopy cover2 / 0.024 / 1 / 4.07 / 0.045
Presence of a ditch / 0.062 / 1 / 10.69 / 0.001
Error / 165

Figure S5NMDS ordination diagram displaying axes 1 and 3 based on species composition in different habitats using Bray-Curtis distance. Final solution comprised of four dimensions with a stress of 0.15. Ellipses denote 50% confidence ellipses of habitat types. Rare species represent species that were recorded in fewer than three sites