PoPS Global is a species-agnostic network modeling framework that couples international trade networks with core drivers of biological invasions using open, globally available databases to forecast pest introductions and global spread through bridgehead populations.
PoPS Global is a spatio-temporal stochastic network modeling approach wherein network nodes represent geographical areas (e.g., countries, regions, ports) and bidirectional network edges (i.e., connections between nodes) represent the movement of goods via trade pathways. Potential plant pest import and export is modeled along these pathways by integrating global trade data, pest occurrence, host species distribution, and climate conditions. The model predicts the probability of introduction (i.e., successful entry and establishment) for every node in the network at each time step. Nodes with successful introductions then become bridgehead populations with the potential for transmitting the pest in the subsequent time step, or after an optional latency period.
The model consists of three equations calculating separate but related probabilities: 1) entry, 2) establishment, and 3) introduction. These terms align with definitions used by the United States Department of Agriculture Animal and Plant Health Inspection Service (USDA APHIS) and correspond, respectively, to transport, introduction, and establishment as defined by Blackburn et al. (2011).
Probability of entry captures processes controlling movement between globally distributed nodes. It is a function of the amount of traded goods capable of transporting the pest, the likelihood of a pest surviving the journey, and, optionally, the phytosanitary capacity of importing and exporting countries.
Probability of establishment captures conditions and ecological processes within a node area. Establishment probability increases with environmental suitability, which is modeled as a Gaussian function of the climate dissimilarity between the two trading nodes, and percent area without host species in the destination node. Optionally, the probability can be adjusted by the pest’s ability to survive on multiple hosts (e.g., number of host taxonomic families weighted by phylogenetic diversity of hosts).
Probability of introduction is a function of the probability of entry (inter-node processes) and the probability of establishment (intra-node processes), and is used in a binomial distribution to determine if a successful introduction occurs.
The PoPS Global workflow is implemented in a series of Jupyter Notebooks for acquiring and formatting the model input data and running the model. To use the data acquisition and formatting notebook, the user must provide as input a raster of the Köppen-Geiger Climate Classification (Beck et al., 2018), a comma-separated values file of phytosanitary capacity scores (if applicable), a global, binary raster of host presence and absence, and an environmental file with information on where to store the model outputs. All other data are acquired and formatted within the notebook workflow. Another notebook is used to configure the desired model parameters, scenario configurations, and number of iterations. The model is run within the notebook and the results are saved locally. Additional notebooks are also available for reading the model outputs and creating result summaries and plots.
We are using Pipenv for our virtual environment. To use, install on your system:
pip install pipenv
To run use
pipenv install
pipenv shell
To install a new package
pipenv install "some package"
- Chelsey Walden-Schreiner, NCSU Center for Geospatial Analytics
- Chris Jones, NCSU Center for Geospatial Analytics
- Kellyn P. Montgomery, NCSU Center for Geospatial Analytics
- Ariel Saffer, NCSU Center for Geospatial Analytics
- Vaclav Petras, NCSU Center for Geospatial Analytics
- Ben Seliger, NCSU Center for Geospatial Analytics
The simulation code is open source under GNU GPL >=v2 (see the LICENSE file for details).
This research is funded by USDA APHIS. The findings do not necessarily represent the views of USDA APHIS.
Please note that this is a simulation and it needs to be calibrated to give any realistic or actionable results. Results presented here are examples for demonstration purposes only.