Skip to main content
Dryad logo

The role of dispersal limitation and reforestation in shaping the distributional shift of a forest herb under climate change

Citation

Van Daele, Frederik; Honnay, Olivier; De Kort, Hanne (2021), The role of dispersal limitation and reforestation in shaping the distributional shift of a forest herb under climate change, Dryad, Dataset, https://doi.org/10.5061/dryad.msbcc2fxx

Abstract

Aim: Forest herbs might be unable to track shifts in habitat suitability due to rapid climate change and habitat fragmentation. In this study, we quantified the role of dispersal limitation and the potential mitigating effect of large-scale reforestation on the redistribution of the herbaceous forest plant species Primula elatior under climate change.

Location: Europe

Methods: High resolution (100 m) landscape-scale and macro-climatic variables were combined to predict range-wide habitat suitability using Maxent. Dispersal limitation was modelled, based on isolation-by-resistance (IBR) principles through integration of circuit theory and genomic data, to assess patch accessibility and metapopulation stability under climate change. Large scale reforestation was evaluated as a potential mitigating strategy by incorporating a land-use change scenario into the distribution and dispersal models.

Results: Landscape-scale variables contributed significantly to the distribution of P. elatior (78.33%) and to the accuracy of our model (AUC=0.81). Isolation-by-resistance (R²cond=0.92) was driven by land-use (45.5%), distance from rivers (36.4%) and elevation (18.2%). It was estimated that 46.4±13.9% (mean±SD of climate change scenarios) of the total distribution area would be lost due to climate change by 2050 and an additional 15.6±1.7% (mean±SD) of the distribution would not be accessible through migration. The median latitude of the patch distribution shifted 183.2±34.8 km (mean±SD) northwards and 58.1±9.3 km (mean±SD) to more maritime regions. The patch accessibility was low in these regions and the metapopulation stability decreased considerably in the south of the distribution. Reforestation mitigated 54.1±18.2% (mean±SD) of the accessible distribution area loss and 49.5±4.2% (mean±SD) of the decrease in metapopulation stability.

Main conclusion: To alleviate the loss of the accessible distribution area of P. elatior under climate change, it will be required to integrate climate mitigation strategies (RCP 2.6), range-wide afforestation, restoration of ecological connectivity and focused assisted migration to newly available habitat.

Methods

Data collection and processing

Input_data_SDM.xlsx

Training and validation data (occurrences and pseudo-absences) used for the species distribution modelling. The data processing methodology can be found in section “Species occurrence data and pseudo-absence generation” of the manuscript. Coordinates of the occurrences were removed due to contractual obligations. The coordinates can be requested at the agencies listed in table I of Appendix S1 of the manuscript and an overview of the coordinates can be found in Fig. 1 of Appendix S1. The corresponding raster stack values (explanatory variables) of occurrences and pseudo-absences are integrated in this dataset.

PREDMEAN Raster tif files

The methods regarding the projections in space and time of the species distribution models can be found in the section “Model building and projections” of the manuscript. Further info on the input data preparation can be found in Appendix S1-SIII and more in depth evaluation of the models can be found in Appendix SVI-SV.

isolation_by_resistance.zip

Methods corresponding to the isolation by resistance relationship can be found in the section “Modelling dispersal and migration events to determine the accessible area” of the manuscript. Detailed methods about the genome sequencing can be found in Appendix SVII (genetic distance) and detailed methods about the effective resistances can be found in Appendix SVIII.

habitat_patch_shapefiles.zip

Methods in regard to the habitat patch delineation can be found in the “Model building and projections” section of the manuscript, methods regarding the dispersal and migration effective resistances and potential can be found in the “Modelling dispersal and migration events to determine the accessible area” section of the manuscript, and methods regarding the proximity resistance index can be found in the “Assessing shifts in patch distribution and configuration” section of the manuscript. A more in detail description of the genetic optimization procedure and resulting statistics of dispersal potential thresholds can be found in Appendix SVII.

Usage Notes

Datasets

Input_data_SDM.xlsx

    • Input_data_SDM.xlsx
      • Training and evaluation data for the species distribution models

00_CURRENT_PREDMEAN.tif

    • The current projected species distribution for Primula elatior.

01_FUT26_PREDMEAN.tif

    • The projected species distribution for Primula elatior in 2050 under the RCP 2.6 scenario.

01b_FUT26_LUC_PREDMEAN.tif

    • The projected species distribution for Primula elatior in 2050 under the RCP 2.6 scenario with projected large scale reforestation.

02_FUT45_PREDMEAN.tif

    • The projected species distribution for Primula elatior in 2050 under the RCP 4.5 scenario.

02b_FUT45_LUC_PREDMEAN.tif

    • The projected species distribution for Primula elatior in 2050 under the RCP 4.5 scenario with projected large scale reforestation.

03_FUT85_PREDMEAN.tif

    • The projected species distribution for Primula elatior in 2050 under the RCP 8.5 scenario.

03b_FUT85_LUC_PREDMEAN.tif

    • The projected species distribution for Primula elatior in 2050 under the RCP 8.5 scenario with projected large scale reforestation.

isolation_by_resistance.zip

    • genetic_samples_locations.shp
      • Shapefile with the locations of the genetic samples
    • isolation_by_resistance_data.txt
      • text file with genetic distances, effective resistances and their derivatives between pairs of sampling locations
    • isolation_by_resistance_NLS_model.rds
      • The Nonlinear Least Squares model of the isolation by resistance relationship. The rds data format is a R storage data format that can be read with the readRDS(file = "isolation_by_resistance_NLS_model.rds") function.

habitat_patch_shapefiles.zip

    • 00_CURRENT.shp
      • Shapefiles of the current projected habitat patch distribution of Primula elatior.
    • 01_FUT26.shp
      • Shapefiles of the projected habitat patch distribution of Primula elatior in 2050 under the RCP 2.6 scenario.
    • 01b_FUT26_LUC.shp
      • Shapefiles of the projected habitat patch distribution of Primula elatior in 2050 under the RCP 2.6 scenario with projected large scale reforestation.
    • 02_FUT45.shp
      • Shapefiles of the projected habitat patch distribution of Primula elatior in 2050 under the RCP 4.5 scenario.
    • 02b_FUT45_LUC.shp
      • Shapefiles of the projected habitat patch distribution of Primula elatior in 2050 under the RCP 4.5 scenario with projected large scale reforestation.
    • 03_FUT85.shp
      • Shapefiles of the projected habitat patch distribution of Primula elatior in 2050 under the RCP 8.5 scenario.
    • 03b_FUT85_LUC.shp
      • Shapefiles of the projected habitat patch distribution of Primula elatior in 2050 under the RCP 8.5 scenario with projected large scale reforestation.

Metadata

Input_data_SDM.xlsx

Table 1: Field names and their description of the input data for the species distribution models. Coordinates for the occurrence data was removed due to contractual obligations.

Field

Description

ID

Occurrence ID. Concatenated field name based on the original dataset and occurrence number.

aspect

Raster value extract of aspect (exposure)

carpinus

Raster value extract of landscape-scale Carpinus tree density

CLC

Raster value extract corine land-use

DEM

Raster value extract of the digital elevation model in meters

dry_season_18

Raster value extract of precipitation in the dry season (worldclim 18)

fagus

Raster value extract of landscape-scale Fagus tree density

quercus

Raster value extract of landscape-scale Quercus tree density

River_log

Raster value extract of logarithmic river distance

seasonality_4

Raster value extract of temperature seasonality (worldclim 4)

Soil_physical_PC1

Raster value extract of the first principal component of physical soil parameters

Soil_physical_PC2

Raster value extract of the second principal component of physical soil parameters

wet_month_13

Raster value extract of precipitation in the wetterst month (worldclim 13)

WPI

Raster value extract of the wetness proximity index

longitude

X coordinate in LAEA89

latitude

Y coordinate in LAEA89

dataset

Dataset name: training or evaluation / occurrence or pseudo-absence / repetition ID

 

Table 2: different datasets combined as determined by the dataset field. Each dataset name has 7 repetitions equal to the model repetition.

dataset names

Description

01_occurrence_training

Spatially thinned Primula elatior occurrences from 2000-2020

02_pseudoabsence_training

Spatially thinned background data (pseudo-absences) with approx. 200% quantity compared to occurrences

03_occurrence_evaluation

Spatially thinned and geographically independent evaluation data of Primula elatior from 2000-2020

04_pseudoabsence_evaluation

Spatially thinned background data (pseudo-absences) for evaluation with approx. 200% quantity compared to occurrences

 

PREDMEAN Raster tif files

  • All raster predictins indicate the average occurrence probability ranging between 0 and 1 based on the 7 repetitions.

isolation_by_resistance.zip

Table 3: table of the isolation_by_resistance metadata. Source depicts the file that is described, field describes the column name and description describes the field name.

Source

Field

Description

isolation_by_resistance_data

to

ID of samples habitat patches dispersal target (bidirectional matrix)

isolation_by_resistance_data

from

ID of samples habitat patches source target (bidirectional matrix)

isolation_by_resistance_data

NEI_dist

Nei's genetic distance

isolation_by_resistance_data

resistance

Effective resistance (based on the optimized resistance matrix)

isolation_by_resistance_data

inverse_standardised

_genetic_distance

inverse standardised genetic distance (transformed Nei's distance)

isolation_by_resistance_data

logaritmic_resistance

Logarithmic effective resistance

genetic_samples_locations

POP_ID

ID of samples

genetic_samples_locations

X

LAEA89 coordinate in meters

genetic_samples_locations

Y

LAEA89 coordinate in meters

isolation_by_resistance_NLS

_model.rds

m

Contains NLS model in rds R data format based on the isolation_by_resistance_data

 

habitat_patch_shapefiles.zip

Table 4: Description of Primula elatior habitat patch metrics in the shapefiles. “Field name” is the ID of each field, “Unit” describes the unit of the metric and “description” describes the metric. .

Field name

Unit

Description

ID

positive integer

Patch ID

 X

meters (LAEA89)

X coordinate of patch centroid

 Y

meters (LAEA89)

Y coordinate of patch centroid

 sea_dist

meters

Distance from the sea

 DEM

meters

Elevation

 aspect

degrees

Orientation of slope

 area

Patch area

 area_ha

hectares

Patch area

 NEAR_DIS

meters

Nearest neighbour distance between patches within a scenario

 ERD_SS

effective resistance

Effective resistance of dispersal between nearest neightbour patches with short circuit methodology (no resistance within patch)

 ERD_CEN

effective resistance

Effective resistance of dispersal from patch centroid to its nearest neighbour (with additional resistance within the patch)

 ERD_WPD

effective resistance

Effective resistance of dispersal from patch centroid to patch edge (within-patch dispersal): res_cen - res_SS

 M_DIS

meters

Nearest neighbour distance between current patches and projected patches in 2050

 MCL_DIS

meters

Nearest neighbour distance between current source patches and projected patches in 2050 (source patches with a within-patch dispersal potential > 50)

 ER_M

effective resistance

Effective resistance of migration between current patches and its nearest neighbour projected future patch

 ER_M_WL

effective resistance

Effective resistance of migration between current source patches and nearest neighbour projected future patches (source patches with a within-patch dispersal potential > 50)

 POT_DSS

inverse standardised genetic distance

Dispersal potential (predicted inverse standardized genetic distance)

 POT_WPD

inverse standardised genetic distance

Within-patch dispersal potential  (predicted inverse standardized genetic distance)

 POT_M

inverse standardised genetic distance

Migration potential from all patches (predicted inverse standardized genetic distance)

 POT_M_WL

inverse standardised genetic distance

Migration potential from source patches (predicted inverse standardized genetic distance used for migration potential in the manuscript)

 PRI

sum(area/resistance)

Proximity resistance index with a distance threshold of 5 km

Funding

Fonds Wetenschappelijk Onderzoek, Award: G091419N