rconnect

rconnect is a simple package that implements riverscape connectivity measures for quantifying the colonization potential of wind dispersed plants species as described by Wagner & Wöllner (2023):

effectiveSeedrain
effectiveConnections
effectiveDistance
colonizationPotential
absoluteConnections
connectionCapacity

It further provides a helper function to create a negative exponential dispersal kernel, functions to create a connectivity and distance matrix based on effectiveConnections and effectiveDistance and a simple example raster file with suitable habitats.

Installation

You can install the current development version of rconnect from GitHub with:

# install.packages("devtools")
devtools::install_github("TCWagner/rconnect")

If you use RStudio, you can just download the package as .tgz and install it through Packages - Install - Install from: Package Archive File.

Example

Here is a basic example how to work with the package and calculate simple connectivity metrics of a riverscape.

Let’s find out how well the habitats of a riverscape (a riverine landscape) are connected for a certain species. As example we chose the Asteracea Chondrilla chondrilloides, an wind dispersed species well adapted to open gravel bars of alpine rivers. For details about the species ecology, dispersal and status, see Wöllner et al. 2022.

We start with a raster file, containing the suitable habitats for our species. Suitable habitats need to have a value > 0; 0 codes for unsuitable habitats. Here we use our example data that comes with our package: habitats_lech

library(raster)
#> Loading required package: sp
library(rconnect)

## load the example raster with suitable habitats. suitable habitats need to be coded with values > 0
## you may use your own raster, here we use our example data:
data(habitats_lech)
plot(habitats_lech)

Now we need the dispersal kernel for the species under consideration as a matrix. We can create a simple, negative exponential dispersal kernel with the function dispersalKernel. Assuming a negative exponential decrease of the seeds with distance and a dispersal distance of ~14m the decay is 0.19.

To create the kernel, we need to provide the cell size of our raster containing the suitable habitats (5m), and the intended radius of our kernel in cells. By default, the kernel center cell will be set to 0 and the kernel normalized to sum up to 1.

cckernel <- dispersalKernel(cellsize=5, radius=7, decay=0.19)
#> Loading required namespace: igraph

Lets start to calculate the seed rain that can be exchanged between neighboring habitats (effectiveSeedrain). If we do not summarize, the output will be a raster with the relative contribution that each cell of an habitat has.

eS <- effectiveSeedrain(habitats_lech, cckernel, summarize=F)
plot(eS)

Now, lets see how many effective Connections a patch has with the other patches of the riverscape

eCM <- effectiveConnectionsMatrix(habitats_lech, cckernel)
eCM
#>              [,1]        [,2]      [,3]       [,4]       [,5]      [,6]
#>  [1,]          NA 0.005021085 0.0000000 0.00000000 0.00000000 0.0000000
#>  [2,] 0.005021085          NA 0.3773102 0.00000000 0.00000000 0.0000000
#>  [3,] 0.000000000 0.377310226        NA 0.06423960 1.45851127 0.0000000
#>  [4,] 0.000000000 0.000000000 0.0642396         NA 0.04449116 0.0000000
#>  [5,] 0.000000000 0.000000000 1.4585113 0.04449116         NA 0.1585712
#>  [6,] 0.000000000 0.000000000 0.0000000 0.00000000 0.15857125        NA
#>  [7,] 0.000000000 0.000000000 0.0000000 0.00000000 0.36850376 0.0000000
#>  [8,] 0.000000000 0.000000000 0.0000000 0.00000000 0.98020300 0.0000000
#>  [9,] 0.000000000 0.000000000 0.0000000 0.00000000 0.03742227 0.0000000
#>            [,7]     [,8]       [,9]
#>  [1,] 0.0000000 0.000000 0.00000000
#>  [2,] 0.0000000 0.000000 0.00000000
#>  [3,] 0.0000000 0.000000 0.00000000
#>  [4,] 0.0000000 0.000000 0.00000000
#>  [5,] 0.3685038 0.980203 0.03742227
#>  [6,] 0.0000000 0.000000 0.00000000
#>  [7,]        NA 0.000000 0.00000000
#>  [8,] 0.0000000       NA 0.00000000
#>  [9,] 0.0000000 0.000000         NA

However, if we want to have the total effective Connections or eC that a patch has with all neighbors (that is the connections weighted by distance) we can use:

eC <- effectiveConnections(habitats_lech, cckernel)
eC
#>   patch         eC
#> 1     1 0.00000000
#> 2     2 0.37731023
#> 3     3 1.90006110
#> 4     4 0.10873076
#> 5     5 3.04770271
#> 6     6 0.15857125
#> 7     7 0.36850376
#> 8     8 0.98020300
#> 9     9 0.03742227

We may want to have this a a raster file that we can use for further modeling. So lets set summarize to FALSE:

eCr <- effectiveConnections(habitats_lech, cckernel, summarize=F)
plot(eCr)

Similarly, we can calculate the effective Distance for each patch (eD):

eD <- effectiveDistance(habitats_lech, cckernel)
eD
#>   patch          eD
#> 1     1 25.12378155
#> 2     2  4.56275168
#> 3     3  0.00000000
#> 4     4 10.52994343
#> 5     5  0.00000000
#> 6     6  8.73928558
#> 7     7  4.73756397
#> 8     8  0.09489125
#> 9     9 15.59165370

We may wish to have your results a raster file for further analysis instead of a simple table. No problem, we just call the respective functions with summarize=FALSE option and will get a raster where each patch is assigned the respective value:

eDr <- effectiveDistance(habitats_lech, cckernel, summarize=FALSE)
plot(eDr)

Finally we want a summary of our metrics and the total colonization Potential of our riverscape:

cP <- colonizationPotential(habitats_lech, cckernel)
cP
#>              habitats threshold        cP   cP_rsd       eD  eDm_rsd nCm
#> 1 cc_clumped_lech_ehd         0 0.7765052 1.347085 7.708875 1.094959   2
#>     nCm_rsd       cCm  cCm_rsd
#> 1 0.8291562 0.3262841 1.007004

So, our riverscape has a colonization Potential of 0.776 (~78%) for Chondrilla chondrilloides. The average effective Distance a patch has is 7.7m, clearly within the species dispersal distance. On average, each patch has 2 connections with a capacity of 0.33. However, the relative standard deviation of all parameters is quite high, indicating an unequal distribution of the connectivity. Note, that the average effective Distance (eDm) does only consider connected patches. nCm gives us the average number of Connections a patch has, and cCm tells us the average connection Capacity (relative amount of seeds) that can be exchanged by a connection.

Final Note

The basic functions of this package, effectiveConnections and effectiveSeedrain provide spatially explicit data if needed. Though currently we do neither consider the actual occupancy of habitats or barriers, both can easily be combined with our functions. The respective data can be used for further modelling. We do not include long distance dispersal here, because here the mechanisms are different and more complex. A package suitable for this is currently under development.

References

Woellner, R., Bräuchler, C., Kollmann, J., & Wagner, T. C. (2022). Biological Flora of Central Europe: Chondrilla chondrilloides (Ard.) H. Karst. Perspectives in Plant Ecology, Evolution and Systematics, 54, 125657.

Wagner, T. C., Woellner, R. (2023). A new set of metrics and framework to assess the colonization potential of riverscapes by wind-dispersed plant species. Scientific Reports 13, 20097. http://dx.doi.org/10.1038/s41598-023-47477-y