• 1 SpeciesPoolR
• 2 Motivation for the pacakge
  • 2.1 Rare species are common and important
• 3 Required Data Files
  • 3.1 Species List File
  • 3.2 Shapefile
  • 3.3 Raster Template File
  • 3.4 Land-Use Raster File
  • 3.5 Land-Use Suitability Raster File
• 4 Using SpeciesPoolR Manually
  • 4.1 Importing and Downloading Species Presences
  • 4.2 Creating Spatial Buffers and Habitat Filtering
  • 4.3 Generating summary biodiversity statistics
• 5 Running the SpeciesPoolR Workflow
  • 5.1 How It Works
• 6 References

1 SpeciesPoolR

The goal of the SpeciesPoolR package is to generate potential species pools and their summary metrics in a spatial way. You can install the package directly from GitHub:

#install.packages("remotes")
remotes::install_github("derek-corcoran-barrios/SpeciesPoolR")

No you can load the package

library(SpeciesPoolR)

2 Motivation for the pacakge

2.1 Rare species are common and important

In ecological research, the debate on whether rare species outnumber common species within communities is pivotal for understanding biodiversity and guiding conservation efforts. Numerous studies have shown that rare species typically dominate large ecological assemblages, although common species often exert a more substantial influence on overall species richness patterns (Magurran and Henderson 2003; Bregović, Fišer, and Zagmajster 2019; Schalkwyk, Pryke, and Samways 2019). This complexity underscores the need for innovative approaches in studying biodiversity, particularly since rare species are challenging to model using traditional Species Distribution Models (SDMs) due to their low occurrence rates (Boyd et al. 2022).

Given the limitations of SDMs in capturing the dynamics of rare species, it is essential to develop alternative methods for integrating these species into biodiversity assessments and conservation planning. Although rare species contribute uniquely to functional diversity and ecosystem stability, especially in specific habitats (Chapman, Tunnicliffe, and Bates 2018; Säterberg et al. 2019), their elusiveness in ecological models presents a significant challenge. The question of the minimum number of presence records required for reliable SDMs is crucial. Research has shown that while as few as 10-15 presence observations can produce nonrandom models for some species (Støa et al. 2019), others require higher thresholds—ranging from 14 to 25 records depending on the species’ prevalence and geographic range (Proosdij et al. 2016; Sampaio and Cavalcante 2023). These findings suggest that even sparse datasets can be useful, but the threshold varies significantly depending on species traits and habitat characteristics. Therefore, researchers must explore novel analytical frameworks and conservation strategies that better accommodate the ecological importance of rare species, thereby enhancing our ability to manage and preserve biodiversity effectively (Reddin, Bothwell, and Lennon 2015).

In highly degraded habitats, such as Denmark, where over 60% of the land is dominated by agriculture and less than 10% remains as natural habitat, traditional SDMs may face further limitations. The scarcity of natural habitats means that presence records are often skewed towards human-modified landscapes, complicating the modeling of species’ ecological preferences. In such contexts, where the majority of occurrences may not reflect the species’ natural behaviors or habitat use, relying on complex SDMs could lead to misleading predictions. Instead, simpler algorithms that incorporate basic dispersal mechanisms and habitat filtering might be more effective. By reducing assumptions about habitat preferences, these methods can provide a more realistic framework for conservation planning, particularly when dealing with the restoration of agricultural lands into natural habitats.

For rare species, and indeed for many others, this approach may offer a more practical solution in scenarios where detailed ecological data is sparse or unreliable. Studies have suggested that in such landscapes, simplistic models that prioritize dispersal and broad habitat suitability over intricate ecological niches can better capture species’ potential distributions and their responses to environmental changes (Guisan et al. 2006; Thuiller et al. 2005), an example to this approach would be range bagging (Drake 2015). This pragmatic approach is especially pertinent when planning conservation actions in areas where habitat degradation has left little intact nature, and it ensures that even under data constraints, effective biodiversity management can still be pursued.

3 Required Data Files

To effectively execute the SpeciesPoolR workflow, a set of essential data files must be provided. These files contain the necessary spatial and taxonomic information that underpin the various analytical steps in the package. Below, we detail each required file and its role within the workflow.

3.1 Species List File

• File Type: CSV or Excel file

• Description: The species list file serves as the foundational dataset, comprising the species of interest for your analysis. At a minimum, this file must include a column for the scientific names of species (Species). Additional taxonomic columns, such as Kingdom, Class, and Family, may also be included to facilitate filtering and subgroup analyses.

An example of this file is provided within the package and can be accessed using the following code:

exampleSpecies <- system.file("ex/Species_List.csv", package="SpeciesPoolR")
print(exampleSpecies)
#> [1] "/home/au687614/R/x86_64-pc-linux-gnu-library/4.4/SpeciesPoolR/ex/Species_List.csv"

This dataset is further discussed in the section on Reading and Filtering Data, with a filtered subset displayed in Table 4.1.

3.2 Shapefile

• File Type: Shapefile (.shp)

• Description: The shapefile delineates the geographic area of interest, which can range from a broad region, such as a country, to a more specific locality, such as a nature reserve. This file is utilized to spatially constrain species occurrences, ensuring that only those within the defined boundaries are included in the analysis.

If a shapefile is unavailable, a two-letter country code (e.g., “DK” for Denmark) may be provided as an alternative to specify the area of interest.

An example shapefile is included in the package and can be accessed as follows:

shp <- system.file("ex/Aarhus.shp", package="SpeciesPoolR")
print(shp)
#> [1] "/home/au687614/R/x86_64-pc-linux-gnu-library/4.4/SpeciesPoolR/ex/Aarhus.shp"

The shapefile’s application is illustrated in the section on Counting Species Presences, where it is used to delineate the boundaries of Aarhus commune, as shown in Figure 3.1.

Figure 3.1: Outline of the comune of Aarhus

3.3 Raster Template File

• File Type: Raster file (e.g., .tif)

• Description: The raster template file is employed as a spatial reference for rasterizing species presence buffers. It must cover the entire area of interest and possess a resolution appropriate for the intended analysis. This template ensures consistent spatial alignment across all raster-based operations.

You can explore an example of this file using the following code:

template <- system.file("ex/LU_Aarhus.tif", package="SpeciesPoolR")
print(template)
#> [1] "/home/au687614/R/x86_64-pc-linux-gnu-library/4.4/SpeciesPoolR/ex/LU_Aarhus.tif"

The raster template’s role in buffer creation is further explained in the section on Creating Buffers Around Species Presences, with an example shown in Figure 3.2.

Figure 3.2: Raster of the Aarhus comune, the package will use Non NA cells as part of the template

3.4 Land-Use Raster File

• File Type: Raster file (e.g., .tif)

• Description: This file contains land-use classifications for the study area, where each raster cell is assigned to a specific land-use category (e.g., forest, wetland, urban). This data is crucial for modeling habitat suitability, enabling the filtering of species occurrences based on the prevalent land uses within their potential habitats.

An example file is provided in the package:

LU <- system.file("ex/LU_Aarhus.tif", package="SpeciesPoolR")
print(LU)
#> [1] "/home/au687614/R/x86_64-pc-linux-gnu-library/4.4/SpeciesPoolR/ex/LU_Aarhus.tif"

The land-use raster is identical to the template shown in Figure 3.2.

3.5 Land-Use Suitability Raster File

• File Type: Raster file (e.g., .tif)

• Description: This file comprises binary suitability values for various land-use types within the study area, indicating whether each land-use type is suitable (value = 1) or unsuitable (value = 0) for the habitat of interest. The data is subsequently transformed into a long-format table, which is integral to the habitat filtering and species distribution modeling processes.

An example raster file is available in the package, and its application is discussed in the section on Preparing Land-Use Data. A visualization of this file is presented in Figure 3.3.

Figure 3.3: Landuse suitability for 8 different landuses in the aarhus commune

4 Using SpeciesPoolR Manually

4.1 Importing and Downloading Species Presences

4.1.1 Step 1: Reading and Filtering Data

If you are going to use each of the functions of the SpeciesPoolR manually and sequentially, the first step would be to read in a species list from either a CSV or an XLSX file. You can use the get_data function for this. The function allows you to filter your data in a dplyr-like style:

f <- system.file("ex/Species_List.csv", package="SpeciesPoolR")
filtered_data <- get_data(
   file = f,
   filter = quote(Kingdom == "Plantae" & 
                    Class == "Magnoliopsida" & 
                    Family == "Fabaceae")
)

This will generate a dataset that can be used subsequently to count species presences and download species data as seen in table 4.1

redlist_2010	Kingdom	Phyllum	Class	Order	Family	Genus	Species
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia sepium
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Genista	Genista tinctoria
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Trifolium	Trifolium vesiculosum
LC	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia sativa
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Lathyrus	Lathyrus latifolius
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Anthyllis	Anthyllis vulneraria
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia sepium
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Lathyrus	Lathyrus japonicus
NA	Plantae	Magnoliophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia villosa

Table 4.1: Species that will be used to generate species pools

4.1.2 Step 2: Taxonomic Harmonization

Next, you should perform taxonomic harmonization to ensure that the species names you use are recognized by the GBIF taxonomic backbone. This can be done using the Clean_Taxa function:

Clean_Species <- SpeciesPoolR::Clean_Taxa(filtered_data$Species)

The resulting data frame, with harmonized species names, is shown in table 4.2

Taxa	matched_name2	confidence	canonicalName	kingdom	phylum	class	order	family	genus	species	rank
Vicia sepium	Vicia sepium	99	Vicia sepium	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia sepium	SPECIES
Genista tinctoria	Genista tinctoria	99	Genista tinctoria	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Genista	Genista tinctoria	SPECIES
Trifolium vesiculosum	Trifolium vesiculosum	99	Trifolium vesiculosum	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Trifolium	Trifolium vesiculosum	SPECIES
Vicia sativa	Vicia sativa	97	Vicia sativa	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia sativa	SPECIES
Lathyrus latifolius	Lathyrus latifolius	98	Lathyrus latifolius	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Lathyrus	Lathyrus latifolius	SPECIES
Anthyllis vulneraria	Anthyllis vulneraria	97	Anthyllis vulneraria	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Anthyllis	Anthyllis vulneraria	SPECIES
Lathyrus japonicus	Lathyrus japonicus	99	Lathyrus japonicus	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Lathyrus	Lathyrus japonicus	SPECIES
Vicia villosa	Vicia villosa	97	Vicia villosa	Plantae	Tracheophyta	Magnoliopsida	Fabales	Fabaceae	Vicia	Vicia villosa	SPECIES

Table 4.2: Taxonomicallty harmonized dataset

4.1.3 Step 3: Counting Species Presences

After harmonizing the species names, it’s important to obtain the number of occurrences of each species in your study area, especially if you plan to calculate rarity. You can do this using the count_presences function. This function allows you to filter occurrences by country or by a shapefile. Below is an example for Denmark:

# Assuming Clean_Species is your data frame
Count_DK <- count_presences(Clean_Species, country = "DK")

The resulting data frame of species presences in Denmark is shown in table 4.3

knitr::kable(Count_DK, caption = "Counts of presences for the different species within Denmark")

family	genus	species	N
Fabaceae	Vicia	Vicia sepium	2901
Fabaceae	Genista	Genista tinctoria	988
Fabaceae	Trifolium	Trifolium vesiculosum	0
Fabaceae	Vicia	Vicia sativa	17380
Fabaceae	Lathyrus	Lathyrus latifolius	685
Fabaceae	Anthyllis	Anthyllis vulneraria	8880
Fabaceae	Lathyrus	Lathyrus japonicus	3905
Fabaceae	Vicia	Vicia villosa	243

Table 4.3: Counts of presences for the different species within Denmark

Alternatively, you can filter by a specific region using a shapefile. For example, to count species presences within Aarhus commune:

shp <- system.file("ex/Aarhus.shp", package="SpeciesPoolR")

Count_Aarhus <- count_presences(Clean_Species, shapefile = shp)

The resulting data.frame for Aarhus commune is shown int table 4.4

family	genus	species	N
Fabaceae	Vicia	Vicia sepium	283
Fabaceae	Genista	Genista tinctoria	27
Fabaceae	Trifolium	Trifolium vesiculosum	0
Fabaceae	Vicia	Vicia sativa	467
Fabaceae	Lathyrus	Lathyrus latifolius	41
Fabaceae	Anthyllis	Anthyllis vulneraria	153
Fabaceae	Lathyrus	Lathyrus japonicus	39
Fabaceae	Vicia	Vicia villosa	10

Table 4.4: Counts of presences for the different species within Aarhus commune

Now it is recommended to eliminate species that have no occurrences in the area, this is done automatically in the workflow version:

library(data.table)
Count_Aarhus <- Count_Aarhus[N > 0,]

So that then we can retrieve the species presences using the function SpeciesPoolR::get_presences.

Presences <- get_presences(species = Count_Aarhus$species, shapefile = shp)
#> [1] "Geometry created: POLYGON ((10.401438 56.302419, 10.048024 56.355225, 9.886316 56.019928, 10.239729 55.966657, 10.401438 56.302419))"

there we end up with 1077 presences for our 7 species.

4.2 Creating Spatial Buffers and Habitat Filtering

4.2.1 Step 1 Creating Buffers Around Species Presences

Once you have identified the species presences within your area of interest, the next step is to create spatial buffers around these occurrences. These buffers represent the potential dispersal range of each species, helping to assess areas where the species might establish itself given a specified dispersal distance.

To create these buffers, you’ll use a raster file as a template to rasterize the buffers and specify the distance (in meters) representing the species’ dispersal range.

Raster <- system.file("ex/LU_Aarhus.tif", package="SpeciesPoolR")

buffer500 <- make_buffer_rasterized(Presences, file = Raster, dist = 500)

In this example, the make_buffer_rasterized function generates a 500-meter buffer around each occurrence point in the Presences dataset. The function utilizes the provided raster file as a template for rasterizing these buffers.

The resulting buffer500 data frame indicates which raster cells are covered by the buffer for each species. Table 4.5 displays the first 10 observations of this data frame, providing a detailed view of the buffer’s overlap with raster cells, listing each cell and the corresponding species within that buffer.

cell	species
26	Vicia sepium
27	Vicia sepium
28	Vicia sepium
29	Vicia sepium
30	Vicia sepium
161	Vicia sepium
162	Vicia sepium
163	Vicia sepium
164	Vicia sepium
165	Vicia sepium

Table 4.5: Raster cells within the 500-meter buffer of each species

This table provides a detailed view of how the buffer overlaps with the raster cells, listing each cell and the corresponding species present within that buffer.

4.2.2 Step 2: Habitat Filtering

After creating the buffers, the next logical step is to filter these areas based on habitat suitability. This allows you to focus on specific land-use types or habitats where the species is more likely to thrive. Habitat filtering typically involves using raster data to refine or subset the buffer areas according to the desired habitat criteria.

4.2.2.1 Preparing Land-Use Data

Before you can apply habitat filtering, you need to prepare a long-format land-use table that matches each raster cell to its corresponding habitat types. This is done using the generate_long_landuse_table function, which takes the path to your raster file and transforms it into a long-format data frame. The function also filters the data to include only those cells where the suitability value is 1 for at least one land-use type.

# Get path for habitat suitability
HabSut <- system.file("ex/HabSut.tif", package = "SpeciesPoolR")

# Generate the long-format land-use table
long_LU_table <- generate_long_landuse_table(path = HabSut)

This is crucial for the next steps, the result is shown in table 4.6, as it links each raster cell to potential habitats, enabling you to match species occurrences to suitable environments within their buffer zones.

cell	Habitat
79	OpenDryPoor
80	OpenDryPoor
81	OpenDryPoor
82	OpenDryPoor
83	OpenDryPoor
214	OpenDryPoor
215	OpenDryPoor
216	OpenDryPoor
217	OpenDryPoor
218	OpenDryPoor

Table 4.6: First 10 observations of landuse suitability per cell

4.2.2.2 Applying Habitat Filtering

Once you have the long-format land-use table, you can proceed with habitat filtering. To achieve this, you’ll use the ModelAndPredictFunc, which takes the presence data frame (e.g., Presences) obtained through the get_presences function and the land-use raster. This comprehensive function encompasses several critical steps:

1- Grouping Data by Species: The presence data is grouped by species using group_split, ensuring that each species is modeled individually.

2- Sampling Land-Use Data: For each species, land-use data is sampled at the presence points using the SampleLanduse function.

3- Sampling Background Data: Background points are also sampled from the same land-use raster, providing a contrast to the presence data.

4- Modeling Habitat Suitability: The presence and background data are combined and passed to the ModelSpecies function. This function fits a MaxEnt model to predict habitat suitability across the different land-use types.

5- Predicting Suitability: The fitted model is then used to predict habitat suitability for each species across all available land-use types.

Habitats <- ModelAndPredictFunc(DF = Presences, file = Raster)

The resulting Habitats data frame contains continuous suitability predictions for each species across various land-use types. Table 4.7 shows the first 9 observations, illustrating the predicted habitat suitability scores for the first species in each land-use type.

knitr::kable(Habitats[1:9,], caption = "Predicted habitat suitability scores across various land-use types for the first species. The values represent continuous predictions, indicating the relative likelihood of species presence in each land-use category.")

Landuse	Pred	species
OpenDryRich	1.0000000	Anthyllis vulneraria
OpenDryPoor	1.0000000	Anthyllis vulneraria
ForestWetRich	0.6749865	Anthyllis vulneraria
OpenWetRich	0.6749865	Anthyllis vulneraria
OpenWetPoor	0.6749865	Anthyllis vulneraria
Exclude	0.5127553	Anthyllis vulneraria
ForestDryRich	0.2987050	Anthyllis vulneraria
ForestDryPoor	0.2528306	Anthyllis vulneraria
Exclude	0.6335459	Genista tinctoria

Table 4.7: Predicted habitat suitability scores across various land-use types for the first species. The values represent continuous predictions, indicating the relative likelihood of species presence in each land-use category.

4.2.2.3 Maximum Entropy Model for Land Use Selection

To assess the land use preferences of the target species, we employed a customized Maximum Entropy (MaxEnt) model. This model was developed not as a traditional Species Distribution Model (SDM), but as a lookup table to evaluate the relative selection of land use types. We began by identifying species presence locations and extracting the corresponding land use categories. A minimum convex polygon (MCP) was constructed around these presence points and expanded by 15% to define the background area. Random sampling within this background area provided a distribution of land use types for comparison. The MaxEnt model was then built to evaluate the ratio of the probability density of land use types at presence locations (P(z)) to that at background locations (Q(z)). This ratio, P(z)/Q(z), allowed us to identify land use types that were disproportionately selected by the species. The model utilized a cloglog transformation to convert these ratios into probabilities, providing insights into the species’ land use preferences. The final output was a predictive model indicating the probability of land use selection by the species, highlighting areas where the observed land use differed from that expected based on background availability.

4.2.3 Step 3: Generating Habitat Suitability Thresholds

While continuous predictions provide a detailed picture of habitat suitability, it is often useful to classify these predictions into binary suitability thresholds. Thresholds can help determine areas where species presence is more likely or unlikely based on habitat preferences.

The create_thresholds function facilitates this by generating thresholds based on the modeled land-use preferences, using the 90th, 95th, and 99th percentiles of the predicted suitability values. These thresholds represent the commission rates, helping to define the probability cutoff above which a land-use type is considered suitable for a species.

Here’s how you can generate these thresholds for the species in your dataset:

Thresholds <- create_thresholds(Model = Habitats, reference = Presences, file = Raster)

This will generate de data set with the threshold for the comission rates of 90, 95 and 99th percentile for each species that can be seen in Table 4.8.

species	Thres_99	Thres_95	Thres_90
Anthyllis vulneraria	0.513	0.513	0.513
Genista tinctoria	0.634	0.634	0.634
Lathyrus japonicus	0.407	0.407	0.407
Lathyrus latifolius	0.634	0.634	0.634
Vicia sativa	0.400	0.400	0.400
Vicia sepium	0.299	0.299	0.299
Vicia villosa	0.634	0.634	0.634

Table 4.8: Threshold based on commission rate for the species that are used above

This step produces a data frame containing the thresholds for each species, which can then be used to classify habitat suitability into binary categories, helping you to identify core habitats or areas of higher conservation value.

After we have the continuous thresholds we can generate a lookup table to see which species can inhabit in each landuse type

LookupTable <- Generate_Lookup(Model = Habitats, Thresholds = Thresholds)

This creates Table 4.9, notice how it only shows for each species which habitats are available not the ones that are not.

species	Landuse	Pres
Anthyllis vulneraria	OpenDryRich	1
Anthyllis vulneraria	OpenDryPoor	1
Anthyllis vulneraria	ForestWetRich	1
Anthyllis vulneraria	OpenWetRich	1
Anthyllis vulneraria	OpenWetPoor	1
Lathyrus japonicus	OpenDryPoor	1
Vicia sativa	OpenDryPoor	1
Vicia sativa	OpenDryRich	1
Vicia sativa	OpenWetPoor	1
Vicia sativa	OpenWetRich	1
Vicia sativa	ForestWetRich	1
Vicia sativa	ForestDryRich	1
Vicia sepium	ForestDryRich	1
Vicia sepium	ForestWetRich	1
Vicia sepium	OpenDryPoor	1
Vicia sepium	OpenWetPoor	1
Vicia sepium	OpenWetRich	1
Vicia sepium	OpenDryRich	1

Table 4.9: dummy variable that shows which species can inhabit each habitat type

4.2.4 Step 4: Generating Final Species Presences

In this final step, we apply the make_final_presences function to filter the buffered species presences. This filtering process is done in three stages:

1. Lookup Table Filtering: The function first ensures that each species is only considered in habitats where it can persist based on the species-habitat suitability mappings in the lookup table.

2. Land-Use Table Filtering: Next, it filters these suitable habitats to include only those cells where the specific habitat type could exist, based on the long-format land-use table.

3. Buffer Zone Filtering: Finally, it restricts the potential species occurrences to areas where the species is likely to disperse, as indicated by the spatial buffers generated around species presence points.

The result is a highly refined dataset that specifies, for each species, the exact cells and habitat types where it can potentially occur, combining habitat suitability, land-use distribution, and species dispersal capability.

final_presences <- make_final_presences(Long_LU_table = long_LU_table, 
                                        Long_Buffer_gbif = buffer500,
                                        LookUpTable = LookupTable)

The resulting final_presences table provides detailed information on the potential distribution of each species. It specifies which cells and habitats are suitable for each species, ensuring that only the most plausible locations are considered. In table @(tab:finalpresences), you can see the first 15 observations from this final dataset, which represent the potential habitats where each species could thrive, whereas in table @(tab:summaryfinalpresences), you can see a summary of the number of cells that each species could thrive on each habitat type.

cell	species	Landuse
1018	Anthyllis vulneraria	OpenDryRich
1557	Anthyllis vulneraria	OpenDryRich
1825	Anthyllis vulneraria	OpenDryRich
2093	Anthyllis vulneraria	OpenDryRich
2215	Anthyllis vulneraria	OpenDryRich
2216	Anthyllis vulneraria	OpenDryRich
2218	Anthyllis vulneraria	OpenDryRich
2351	Anthyllis vulneraria	OpenDryRich
2352	Anthyllis vulneraria	OpenDryRich
2353	Anthyllis vulneraria	OpenDryRich
2354	Anthyllis vulneraria	OpenDryRich
2486	Anthyllis vulneraria	OpenDryRich
2488	Anthyllis vulneraria	OpenDryRich
2489	Anthyllis vulneraria	OpenDryRich
2620	Anthyllis vulneraria	OpenDryRich

Table 4.10: First 15 rows of the final presences dataset, showing the cells and land-use types where each species can potentially occur

Landuse	species	N
OpenDryRich	Anthyllis vulneraria	802
ForestWetRich	Anthyllis vulneraria	512
OpenWetRich	Anthyllis vulneraria	512
ForestDryRich	Vicia sepium	254
OpenDryRich	Vicia sepium	254
ForestWetRich	Vicia sepium	141
OpenWetRich	Vicia sepium	141
OpenWetPoor	Anthyllis vulneraria	126
OpenDryPoor	Anthyllis vulneraria	70
OpenDryPoor	Lathyrus japonicus	47
OpenWetPoor	Vicia sepium	31
ForestDryRich	Vicia sativa	26
OpenDryRich	Vicia sativa	26
ForestWetRich	Vicia sativa	18
OpenWetRich	Vicia sativa	18
OpenDryPoor	Vicia sepium	16
OpenWetPoor	Vicia sativa	7

Table 4.11: Summary of number of cells that each species can thrive in for each habitat type

4.3 Generating summary biodiversity statistics

4.3.1 Step 1 Generating Phylogenetic diversity metrics

In order to generate Phylogenetic Diversity measures, the first step is to generate a phylogenetic tree with the species we have, for that we will use the V.Phylomaker package function phylo.makerbased on the megaphylogeny of vascular plants (Jin and Qian 2019; Zanne et al. 2014), this means that we can only use this functions in species pools of plants.

In this case we use the generate_tree from SpeciesPoolR to do so:

tree <- generate_tree(Count_Aarhus)
#> [1] "All species in sp.list are present on tree."

5 Running the SpeciesPoolR Workflow

If you prefer to automate the process and run the SpeciesPoolR workflow as a pipeline, you can use the run_workflow function. This function sets up a targets workflow that sequentially executes the steps for cleaning species data, counting species presences, and performing spatial analysis. This approach is especially useful for larger datasets or when you want to ensure reproducibility.

To run the workflow, you can use the following code. We’ll use the same species filter as before, focusing on the Plantae kingdom, Magnoliopsida class, and Fabaceae family. Additionally, we’ll focus on the Aarhus commune using a shapefile.

shp <- system.file("ex/Aarhus.shp", package = "SpeciesPoolR")
Raster <- system.file("ex/LU_Aarhus.tif", package="SpeciesPoolR")
HabSut <- system.file("ex/HabSut.tif", package = "SpeciesPoolR")


run_workflow(
  file_path = system.file("ex/Species_List.csv", package = "SpeciesPoolR"),
  filter = quote(Kingdom == "Plantae" & Class == "Magnoliopsida" & Family == "Fabaceae"),
  shapefile = shp,
  dist = 500,
  rastertemp = Raster,
  rasterLU = Raster,
  LanduseSuitability = HabSut
)
#> ▶ dispatched target shp
#> ▶ dispatched target Raster
#> ● completed target shp [12.735 seconds]
#> ▶ dispatched target Landuses
#> ● completed target Landuses [0.001 seconds]
#> ▶ dispatched target file
#> ● completed target file [0 seconds]
#> ▶ dispatched target data
#> ● completed target data [1.432 seconds]
#> ▶ dispatched target Clean
#> ● completed target Clean [3.207 seconds]
#> ▶ dispatched branch Count_Presences_33538e94b3809372
#> ● completed branch Count_Presences_33538e94b3809372 [0.493 seconds]
#> ▶ dispatched target Landusesuitability
#> ● completed target Landusesuitability [0.001 seconds]
#> ▶ dispatched target Long_LU_table
#> ● completed target Long_LU_table [0.141 seconds]
#> ▶ dispatched branch Count_Presences_52d72a5ad405e933
#> ● completed branch Count_Presences_52d72a5ad405e933 [0.213 seconds]
#> ▶ dispatched branch Count_Presences_e70f77d9439a4770
#> ● completed branch Count_Presences_e70f77d9439a4770 [0.073 seconds]
#> ▶ dispatched branch Count_Presences_dea4ef8633a449a1
#> ● completed branch Count_Presences_dea4ef8633a449a1 [0.16 seconds]
#> ▶ dispatched branch Count_Presences_69210fc440d13855
#> ● completed branch Count_Presences_69210fc440d13855 [0.142 seconds]
#> ▶ dispatched branch Count_Presences_a61be030e01ebaf5
#> ● completed branch Count_Presences_a61be030e01ebaf5 [0.099 seconds]
#> ▶ dispatched branch Count_Presences_974105e269324d3e
#> ● completed branch Count_Presences_974105e269324d3e [0.068 seconds]
#> ▶ dispatched branch Count_Presences_37d1f8d5f74d852c
#> ● completed branch Count_Presences_37d1f8d5f74d852c [0.066 seconds]
#> ● completed pattern Count_Presences
#> ▶ dispatched target More_than_zero
#> ● completed target More_than_zero [0.004 seconds]
#> ▶ dispatched branch Presences_c112b37cd15959d6
#> ● completed branch Presences_c112b37cd15959d6 [2.178 seconds]
#> ▶ dispatched branch ModelAndPredict_626a53b08dfe709d
#> ● completed target Raster [17.345 seconds]
#> ▶ dispatched branch buffer_626a53b08dfe709d
#> ● completed branch buffer_626a53b08dfe709d [0.282 seconds]
#> ▶ dispatched branch Presences_af64bac105a08467
#> ● completed branch Presences_af64bac105a08467 [1.281 seconds]
#> ▶ dispatched branch buffer_0e19b8cb545404d2
#> ● completed branch buffer_0e19b8cb545404d2 [0.378 seconds]
#> ▶ dispatched branch Presences_daf8d6353bc80f0c
#> ● completed branch Presences_daf8d6353bc80f0c [1.204 seconds]
#> ▶ dispatched branch buffer_edb09c8ec5c9a988
#> ● completed branch buffer_edb09c8ec5c9a988 [0.131 seconds]
#> ▶ dispatched branch Presences_310adeccf6b44725
#> ● completed branch Presences_310adeccf6b44725 [0.574 seconds]
#> ▶ dispatched branch buffer_b226446ac3154351
#> ● completed branch buffer_b226446ac3154351 [0.057 seconds]
#> ▶ dispatched branch Presences_e65f4227e8299cc4
#> ● completed branch Presences_e65f4227e8299cc4 [0.732 seconds]
#> ▶ dispatched branch buffer_0a8436ee3d4f2644
#> ● completed branch buffer_0a8436ee3d4f2644 [0.068 seconds]
#> ▶ dispatched branch Presences_d4b9dc68293bd5b2
#> ● completed branch Presences_d4b9dc68293bd5b2 [0.564 seconds]
#> ▶ dispatched branch buffer_cae8301e59fc4e01
#> ● completed branch buffer_cae8301e59fc4e01 [0.058 seconds]
#> ▶ dispatched branch Presences_88937156c1302a12
#> ● completed branch Presences_88937156c1302a12 [0.432 seconds]
#> ● completed pattern Presences
#> ▶ dispatched branch buffer_a0190cbfdf5f6f1f
#> ● completed branch buffer_a0190cbfdf5f6f1f [0.046 seconds]
#> ● completed pattern buffer
#> ▶ dispatched target Phylo_Tree
#> ● completed branch ModelAndPredict_626a53b08dfe709d [26.624 seconds]
#> ▶ dispatched target rarity_weight
#> ● completed target rarity_weight [0.004 seconds]
#> ▶ dispatched branch ModelAndPredict_0e19b8cb545404d2
#> ● completed branch ModelAndPredict_0e19b8cb545404d2 [2.057 seconds]
#> ▶ dispatched branch ModelAndPredict_edb09c8ec5c9a988
#> ● completed branch ModelAndPredict_edb09c8ec5c9a988 [29.17 seconds]
#> ▶ dispatched branch ModelAndPredict_b226446ac3154351
#> ● completed branch ModelAndPredict_b226446ac3154351 [6.782 seconds]
#> ▶ dispatched branch ModelAndPredict_0a8436ee3d4f2644
#> ● completed target Phylo_Tree [55.56 seconds]
#> ▶ dispatched branch ModelAndPredict_cae8301e59fc4e01
#> ● completed branch ModelAndPredict_cae8301e59fc4e01 [1.408 seconds]
#> ▶ dispatched branch ModelAndPredict_a0190cbfdf5f6f1f
#> ● completed branch ModelAndPredict_a0190cbfdf5f6f1f [1.167 seconds]
#> ● completed branch ModelAndPredict_0a8436ee3d4f2644 [11.749 seconds]
#> ● completed pattern ModelAndPredict
#> ▶ dispatched target Thresholds
#> ● completed target Thresholds [0.648 seconds]
#> ▶ dispatched target LookUpTable
#> ● completed target LookUpTable [0.01 seconds]
#> ▶ dispatched branch Final_Presences_ebf0f62f14548a82
#> ▶ dispatched branch Final_Presences_344cc771c9264c2e
#> ● completed branch Final_Presences_344cc771c9264c2e [0.024 seconds]
#> ▶ dispatched branch Final_Presences_6f1885e07badc469
#> ● completed branch Final_Presences_ebf0f62f14548a82 [0.086 seconds]
#> ▶ dispatched branch Final_Presences_35ecd9eff835718c
#> ● completed branch Final_Presences_6f1885e07badc469 [0.055 seconds]
#> ▶ dispatched branch Final_Presences_af0c167a6a4b9998
#> ● completed branch Final_Presences_35ecd9eff835718c [0.098 seconds]
#> ▶ dispatched branch Final_Presences_5224d468624d4ebb
#> ● completed branch Final_Presences_af0c167a6a4b9998 [0.081 seconds]
#> ▶ dispatched branch Final_Presences_d203d619aa280fd1
#> ● completed branch Final_Presences_5224d468624d4ebb [0.016 seconds]
#> ● completed branch Final_Presences_d203d619aa280fd1 [0.021 seconds]
#> ● completed pattern Final_Presences
#> ▶ dispatched target unique_habitats
#> ▶ dispatched target unique_species
#> ● completed target unique_habitats [0 seconds]
#> ● completed target unique_species [0.006 seconds]
#> ▶ dispatched branch rarity_405e1cf7d36edc08
#> ▶ dispatched branch rarity_77d9a26761e4a007
#> ● completed branch rarity_405e1cf7d36edc08 [0.32 seconds]
#> ▶ dispatched branch output_Rarity_2cf706eb0499c812
#> ● completed branch rarity_77d9a26761e4a007 [0.318 seconds]
#> ▶ dispatched branch output_Rarity_261466b6f238f521
#> ● completed branch output_Rarity_2cf706eb0499c812 [0.227 seconds]
#> ▶ dispatched branch rarity_47ea97700de70215
#> ● completed branch output_Rarity_261466b6f238f521 [0.176 seconds]
#> ▶ dispatched branch rarity_f4a6e9a8f4837219
#> ● completed branch rarity_47ea97700de70215 [0.13 seconds]
#> ▶ dispatched branch output_Rarity_c2db253b33fcf9d9
#> ● completed branch rarity_f4a6e9a8f4837219 [0.231 seconds]
#> ▶ dispatched branch output_Rarity_c1388be691e0be60
#> ● completed branch output_Rarity_c2db253b33fcf9d9 [0.5 seconds]
#> ▶ dispatched branch rarity_bee04486eb86e311
#> ● completed branch output_Rarity_c1388be691e0be60 [0.417 seconds]
#> ▶ dispatched branch rarity_fcb1676d3b2f6824
#> ● completed branch rarity_bee04486eb86e311 [0.099 seconds]
#> ▶ dispatched branch output_Rarity_7d8e043127d47e71
#> ● completed branch rarity_fcb1676d3b2f6824 [0.162 seconds]
#> ● completed pattern rarity
#> ● completed branch output_Rarity_7d8e043127d47e71 [0.096 seconds]
#> ▶ dispatched branch output_Rarity_c8f99e24ce7afc52
#> ▶ dispatched branch PhyloDiversity_405e1cf7d36edc08
#> ● completed branch output_Rarity_c8f99e24ce7afc52 [0.125 seconds]
#> ● completed pattern output_Rarity
#> ▶ dispatched branch PhyloDiversity_77d9a26761e4a007
#> ● completed branch PhyloDiversity_405e1cf7d36edc08 [0.657 seconds]
#> ▶ dispatched branch output_PD_0fed6773594c8b8e
#> ● completed branch output_PD_0fed6773594c8b8e [0.182 seconds]
#> ▶ dispatched branch PhyloDiversity_47ea97700de70215
#> ● completed branch PhyloDiversity_77d9a26761e4a007 [0.766 seconds]
#> ▶ dispatched branch PhyloDiversity_f4a6e9a8f4837219
#> ● completed branch PhyloDiversity_47ea97700de70215 [0.195 seconds]
#> ▶ dispatched branch output_PD_5144605bf3a4ac38
#> ● completed branch output_PD_5144605bf3a4ac38 [0.099 seconds]
#> ▶ dispatched branch PhyloDiversity_bee04486eb86e311
#> ● completed branch PhyloDiversity_bee04486eb86e311 [0.205 seconds]
#> ▶ dispatched branch output_PD_c0ee68bb087b85bd
#> ● completed branch PhyloDiversity_f4a6e9a8f4837219 [0.538 seconds]
#> ▶ dispatched branch output_PD_d34599733447ab6e
#> ● completed branch output_PD_c0ee68bb087b85bd [0.095 seconds]
#> ▶ dispatched branch PhyloDiversity_fcb1676d3b2f6824
#> ● completed branch output_PD_d34599733447ab6e [0.082 seconds]
#> ▶ dispatched branch export_presences_b7bf78e1c1a430c9
#> ● completed branch PhyloDiversity_fcb1676d3b2f6824 [0.679 seconds]
#> ● completed pattern PhyloDiversity
#> ▶ dispatched branch output_PD_32f01d33f95b1de0
#> ● completed branch output_PD_32f01d33f95b1de0 [0.129 seconds]
#> ▶ dispatched branch export_presences_9cb7df6f909cc656
#> ● completed branch export_presences_b7bf78e1c1a430c9 [0.902 seconds]
#> ▶ dispatched branch export_presences_e0501a6e2e4e8857
#> ● completed branch export_presences_e0501a6e2e4e8857 [0.4 seconds]
#> ▶ dispatched branch output_Richness_0fed6773594c8b8e
#> ● completed branch output_Richness_0fed6773594c8b8e [0.102 seconds]
#> ▶ dispatched branch output_PD_9080f2064a157266
#> ● completed branch output_PD_9080f2064a157266 [0.17 seconds]
#> ● completed pattern output_PD
#> ▶ dispatched branch output_Richness_9080f2064a157266
#> ● completed branch output_Richness_9080f2064a157266 [0.201 seconds]
#> ▶ dispatched branch output_Richness_5144605bf3a4ac38
#> ● completed branch output_Richness_5144605bf3a4ac38 [0.122 seconds]
#> ▶ dispatched branch output_Richness_c0ee68bb087b85bd
#> ● completed branch output_Richness_c0ee68bb087b85bd [0.132 seconds]
#> ▶ dispatched branch output_Richness_d34599733447ab6e
#> ● completed branch output_Richness_d34599733447ab6e [0.126 seconds]
#> ▶ dispatched branch output_Richness_32f01d33f95b1de0
#> ● completed branch output_Richness_32f01d33f95b1de0 [0.1 seconds]
#> ● completed pattern output_Richness
#> ● completed branch export_presences_9cb7df6f909cc656 [2.214 seconds]
#> ● completed pattern export_presences
#> ▶ ended pipeline [1.849 minutes]
#> Warning message:
#> 10 targets produced warnings. Run targets::tar_meta(fields = warnings, complete_only = TRUE) for the messages.

5.1 How It Works

The run_workflow function creates a pipeline that:

1. Reads the data from the specified file path.

2. Filters the data using the provided filter expression.

3. Cleans the species names to match the GBIF taxonomic backbone.

4. Counts the species presences within the specified geographic area (in this case, Aarhus).

5. Generates a buffer around the species presences within the specified distance, using a template raster.

6. Prepares the land-use data by generating a long-format table that matches each raster cell to its corresponding habitat types.

7. Predicts habitat suitability for each species across different land-use types using the ModelAndPredictFunc, which models habitat preferences and provides continuous predictions.

8. Generates habitat suitability thresholds for each species based on the predicted suitability scores, using the create_thresholds function to define the 90th, 95th, and 99th percentile thresholds.

9. Builds a lookup table to determine the land-use types each species can inhabit based on the thresholds.

10. Generates the final species presences by filtering the buffered presences according to both the lookup table and the long land-use table, ensuring each species’ potential distribution is consistent with its habitat preferences.

11. Export final presences as csv the final presences of each species and landuse are exported to the folder Field_Final_Presences as csv files.

12. Calculates rarity weights Calculates a Rarity weight for the counts of species that have more than zero presences.

13. Calculates Rarity Calculates and index of relative rarity for each cell on each potential landuse for the cell.

14. Generates a phylogenetic tree for the species in the species list, using the generate_tree function.

15. Calculates phylogenetic phylogenetic diversity for each cell in each potential landuse.

16. Exports phylogenetic phylogenetic diversity, species richness and rarity write out a cloud optimized geotiff of all diversity metrics.

17. Generates a visual representation of the workflow if plot = TRUE.

You can monitor the progress of the workflow and visualize the dependencies between steps using targets::tar_visnetwork(). The result will be similar to running the steps manually but with the added benefits of parallel execution and reproducibility.

This automated approach allows you to streamline your analysis and ensures that all steps are consistently applied to your data. It also makes it easier to rerun the workflow with different parameters or datasets.

The files generated are

#> Results
#> ├── PD
#> │   ├── PD_ForestDryRich.tfw
#> │   ├── PD_ForestDryRich.tif
#> │   ├── PD_ForestWetRich.tfw
#> │   ├── PD_ForestWetRich.tif
#> │   ├── PD_OpenDryPoor.tfw
#> │   ├── PD_OpenDryPoor.tif
#> │   ├── PD_OpenDryRich.tfw
#> │   ├── PD_OpenDryRich.tif
#> │   ├── PD_OpenWetPoor.tfw
#> │   ├── PD_OpenWetPoor.tif
#> │   ├── PD_OpenWetRich.tfw
#> │   └── PD_OpenWetRich.tif
#> ├── Rarity
#> │   ├── Rarity_ForestDryRich.tfw
#> │   ├── Rarity_ForestDryRich.tif
#> │   ├── Rarity_ForestWetRich.tfw
#> │   ├── Rarity_ForestWetRich.tif
#> │   ├── Rarity_OpenDryPoor.tfw
#> │   ├── Rarity_OpenDryPoor.tif
#> │   ├── Rarity_OpenDryRich.tfw
#> │   ├── Rarity_OpenDryRich.tif
#> │   ├── Rarity_OpenWetPoor.tfw
#> │   ├── Rarity_OpenWetPoor.tif
#> │   ├── Rarity_OpenWetRich.tfw
#> │   └── Rarity_OpenWetRich.tif
#> └── Richness
#>     ├── Richness_ForestDryRich.tfw
#>     ├── Richness_ForestDryRich.tif
#>     ├── Richness_ForestWetRich.tfw
#>     ├── Richness_ForestWetRich.tif
#>     ├── Richness_OpenDryPoor.tfw
#>     ├── Richness_OpenDryPoor.tif
#>     ├── Richness_OpenDryRich.tfw
#>     ├── Richness_OpenDryRich.tif
#>     ├── Richness_OpenWetPoor.tfw
#>     ├── Richness_OpenWetPoor.tif
#>     ├── Richness_OpenWetRich.tfw
#>     └── Richness_OpenWetRich.tif

6 References

Boyd, Jennifer Nagel, Jill T. Anderson, Jessica R. Brzyski, Carol J. Baskauf, and J. Cruse-Sanders. 2022. “Eco-Evolutionary Causes and Consequences of Rarity in Plants: A Meta-Analysis.” The New Phytologist. https://doi.org/10.1111/nph.18172.

Bregović, Petra, C. Fišer, and M. Zagmajster. 2019. “Contribution of Rare and Common Species to Subterranean Species Richness Patterns.” Ecology and Evolution 9: 11606–18. https://doi.org/10.1002/ece3.5604.

Chapman, Abbie S. A., V. Tunnicliffe, and A. Bates. 2018. “Both Rare and Common Species Make Unique Contributions to Functional Diversity in an Ecosystem Unaffected by Human Activities.” Diversity and Distributions 24: 568–78. https://doi.org/10.1111/ddi.12712.

Drake, John M. 2015. “Range Bagging: A New Method for Ecological Niche Modelling from Presence-Only Data.” Journal of the Royal Society Interface 12 (107): 20150086.

Guisan, Antoine, Olivier Broennimann, Robin Engler, Mathias Vust, Nigel G Yoccoz, Anthony Lehmann, and Niklaus E Zimmermann. 2006. “Using Niche-Based Models to Improve the Sampling of Rare Species.” Conservation Biology 20 (2): 501–11.

Jin, Yi, and Hong Qian. 2019. “V.PhyloMaker: An r Package That Can Generate Very Large Phylogenies for Vascular Plants.” Ecography 42: 1353–59.

Magurran, A., and P. Henderson. 2003. “Explaining the Excess of Rare Species in Natural Species Abundance Distributions.” Nature 422: 714–16. https://doi.org/10.1038/nature01547.

Proosdij, A. V., M. Sosef, J. Wieringa, and N. Raes. 2016. “Minimum Required Number of Specimen Records to Develop Accurate Species Distribution Models.” Ecography 39: 542–52. https://doi.org/10.1111/ECOG.01509.

Reddin, Carl J., J. Bothwell, and J. Lennon. 2015. “Between-Taxon Matching of Common and Rare Species Richness Patterns.” Global Ecology and Biogeography 24: 1476–86. https://doi.org/10.1111/GEB.12372.

Sampaio, A. C. G., and A. Cavalcante. 2023. “Accurate Species Distribution Models: Minimum Required Number of Specimen Records in the Caatinga Biome.” Anais Da Academia Brasileira de Ciencias 95 2: e20201421. https://doi.org/10.1590/0001-3765202320201421.

Säterberg, Torbjörn, T. Jonsson, J. Yearsley, Sofia Berg, and B. Ebenman. 2019. “A Potential Role for Rare Species in Ecosystem Dynamics.” Scientific Reports 9. https://doi.org/10.1038/s41598-019-47541-6.

Schalkwyk, J., J. Pryke, and M. Samways. 2019. “Contribution of Common Vs. Rare Species to Species Diversity Patterns in Conservation Corridors.” Ecological Indicators. https://doi.org/10.1016/J.ECOLIND.2019.05.014.

Støa, Bente, R. Halvorsen, J. Stokland, and V. I. Gusarov. 2019. “How Much Is Enough? Influence of Number of Presence Observations on the Performance of Species Distribution Models.” Sommerfeltia 39: 1–28. https://doi.org/10.2478/som-2019-0001.

Thuiller, Wilfried, Sandra Lavorel, Miguel B. Araújo, Martin T. Sykes, and I. Colin Prentice. 2005. “Climate Change Threats to Plant Diversity in Europe.” Proceedings of the National Academy of Sciences 102 (23): 8245–50. https://doi.org/10.1073/pnas.0409902102.

Zanne, Amy E., David C. Tank, William K. Cornwell, Jonathan M. Eastman, Stephen A. Smith, Richard G. FitzJohn, Daniel J. McGlinn, et al. 2014. “Three Keys to the Radiation of Angiosperms into Freezing Environments.” American Journal of Botany 506: 89–92.