Migration Mapper: Identifying movement corridors and seasonal ranges for large mammal conservation

Modern tracking technology has facilitated a novel understanding of terrestrial mammal movement while revealing that movements are being truncated and lost. The first step towards conserving mobile animals is identifying movement corridors and key seasonal ranges. Yet, the identification and subsequent mapping of these important areas has remained a challenge due to the analytical skills necessary to conduct such analyses. Migration Mapper (MM) is a user‐friendly software that provides tools to analyse global positioning system (GPS) collar data to create season‐specific, population‐level polygons representing areas where most of a population moves (i.e. movement corridors) and areas where most of a population spends time (e.g. high‐use areas, seasonal ranges). MM consists of six standalone modules including data cleaning and review, seasonal movement delineation, movement model application, calculation of population‐level outputs and visualization of results. Analysis of GPS data using MM can provide the spatial polygons necessary to facilitate conservation and policy planning. New initiatives at the local and global levels are already beginning to use MM to facilitate conservation of large, terrestrial mammals.

MM was to provide field biologists and managers (as well as seasoned data scientists) a pre-packaged but rigorous analytical workflow that removes the need to write code. Furthermore, the ability for users to more easily implement multiple methods within MM provides a robust platform to compare and contrast different space use methods as well as different outputs across taxa and ecosystems (sensu Laver & Kelly, 2008;Lewis et al., 2018). MM consists of a frontend user interface that opens in the user's web browser, where users navigate a straightforward 'point and click' workflow that includes helpful mapping and visualization features. The backend code of MM is scripted in the open-source R program for statistical computing (R Core Team, 2021) and relies on Shiny as the user interface (Chang et al., 2021). All the features in MM are built upon efficient memory management and parallel processing workflows.
The core analytical framework of MM is based upon the methods outlined in Sawyer et al. (2009). Parts of MM were initially released in 2017, and our team provided multiple workshops on how to use MM for biologists, managers and researchers in the western US over subsequent years. Today, MM is being used by numerous biologists, managers and researchers, and has proven to be an important tool to delineate movement corridors and seasonal ranges of ungulates in the western US using a consistent and repeatable framework (Kauffman et al., 2020.

| OVERVIE W OF THE WORKFLOW
MM provides the tools to work from nearly raw GPS collar data to create season-specific, population-level (i.e. merged across individuals) grids and polygons representing areas where most of the population moves (i.e. movement corridors) and areas where most of the population spends significant time (e.g. seasonal ranges, stopover sites). MM is modular and consists of six standalone applications representing distinct steps of the workflow. Modularity clarifies the step-by step process and allows users to easily navigate between steps, fix errors in a previous step or skip steps in the workflow. The modules include (1) data import and review, (2-3) seasonal sequence delineation, (4) movement model application, (5) estimation of population use and corridors and (6) product visualization.
Prior to using MM, users must isolate individuals that form a relatively distinct population (e.g. shared seasonal range). Delineating a distinct population could involve expert opinion, a plotting exercise in a GIS, a spatial clustering analysis (Legendre & Fortin, 1989) or adherence to a policy-based management unit. This step is important because the outputs of MM are strongly dependent on the sample size of a single population and may not be interpretable if two distinct groups of individuals are combined in an analysis (See more details in Appendix S1). Once a distinct population has been identified, users must generate a GPS dataset in ESRI Shapefile format (in any georeferenced coordinate system) that includes a column denoting unique animal ID and the date and time of each GPS location.
To effectively operate MM, users should have an adequate Internet connection (>1 mb/s) to download dependencies and load base maps used in the mapping components. MM requires a host of associated R packages and other dependencies (Appendix S1), which are automatically downloaded and installed when first running MM.
All parameters and options in MM can be adjusted by the user.

| Module 1: Data import and review
Once the user has imported their data, Module 1 checks for erroneous locations (e.g. where there is an unreasonable speed simultaneously approaching and leaving a location) and mortalities (i.e. by identifying strings of locations where the animal did not move a set distance for a set amount of time). MM then presents the user with a mapping tool and some basic plots of the data.
For each individual animal (hereafter ID) and each year of data for each ID (hereafter ID-YR), the user can visualize the points (and lines connecting the points) on an interactive map alongside plots of elevation, displacement from 1 January and speed over time.
The user can interactively review the data and manually identify erroneous locations, mortalities or add comments to each location ( Figure 1, top panel).

| Module 2: Seasonal sequence delineation
In Module 2, the user specifies the biological year start date and how many seasons they would like to potentially isolate for each ID-YR. MM then presents the user with an interactive interface for visualizing each ID-YR of data in space and time with a number of metrics ( Figure 1, bottom panel) inspired by previous studies on how to isolate seasonal movements (Bunnefeld et al., 2011;Cagnacci et al., 2016;Fauchald & Tveraa, 2003;Sawyer et al., 2005;Spitz et al., 2017). Those metrics include Net or Net Squared Displacement (NSD) from the centroid of data from the first week of the biological year, NSD based on 6 months before and after the current biological year, elevation, speed, and multiple scales of first passage time.
The user then manually moves sliders to identify migration periods or other seasons or events of interest (e.g. winter range, parturition period, core summer range, dispersal) for each ID-YR ( Figure 1, bottom panel).

| Module 3: Generating seasonal sequences
Module 3 provides multiple options for generating the seasonal sequences (i.e. distinct strings of GPS locations) for each ID-YR. In addition to generating the sequences for the exact seasons identified in Module 2, users can use the season dates to define additional periods. For instance, if the user used Module 2 to only identify the migratory periods, they could then define the summer period using the end of spring migration and the start of fall migration. It is also possible to define seasonal periods using specified dates, for example defining winter as 1 December through 28 February for every ID-YR. We provide ample flexibility for the user to delineate any biological season of interest, even relatively specific seasonal ranges such as parturition. The line buffer method adds a user-specified spatial buffer to the strait lines between subsequent locations (Figure 2; Appendix S1).

| Module 5: Estimating population use and corridors
A number of steps are necessary to scale from individual sequences to population-level movement corridors and seasonal ranges (Sawyer et al., 2009)

| DISCUSS ION
By providing a user-friendly yet advanced workflow to identify and map movement corridors and important habitats, MM is already helping applied biologists keep pace with the growing volumes of movement data being collected. Contemporary conservation requires a balance between the habitat needs of wide-ranging animals with increased development demands (e.g. housing, agriculture, energy). The maps of movement corridors and important areas produced by MM (see maps in Kauffman et al., 2020) provide science products that managers can use to triage or prioritize key habitats so that impacts on animal populations are avoided or minimized. For example, managers might limit the amount of disturbance allowed in high-use corridors or areas, target them for habitat improvement projects, or prioritize specific private parcels for land protection programmes (Tack et al., 2019). various policy actions across a number of states. For example, the US F I G U R E 1 Migration Mapper's user interface for reviewing and troubleshooting (module 1; top panel) and isolating seasonal sequences (module 2; bottom panel). Global positioning system (GPS) collar data from a migratory mule deer monitored for 2 years are shown as an illustration. Module 1 can be used to initially review data and identify and remove problem points or mortalities by assessing movement metrics on the right hand figures or by clicking directly on individual or groups of points. Module 2 can be used to isolate specific seasons of movement (using maps on left paired with sliders on right), including broad seasonal ranges (e.g. summer range), migration or other movement between seasonal ranges (including nomadic movements), and areas used for parturition. The user identifies how many seasons (or sliders) they want to specify, and once those sliders are moved, the point data in each season are given corresponding, unique colours. The tabs in the bottom right (not shown in figure) provide the user with plots of numerous movement derived metrics (e.g. displacement, elevation, first passage time) over time.

F I G U R E 2
Outline of the steps to calculate population movement corridors and high-use areas (modules 4-5 of migration mapper) from movement sequences isolated using modules 2-3 in migration mapper. Sequences from 25 migratory mule deer monitored with global positioning system (GPS) collars for 2 years are used as an illustration. The movement sequence panel illustrates the steps to calculate occurrence distributions (ODs) and footprints from each isolated sequence of point data from modules 2 and 3. The individual panel illustrates how to merge the multiple occurrences (i.e. across multiple seasons and/or years) for each individual. The bottom two panels illustrate how the footprints or ODs of each individual are merged to create main corridors or high-use areas of the population. Orange arrows represent stages in the workflow where a decision is made or a movement analysis is employed. The workflow related to population high-use areas is relatively flexible and can be used to delineate stopovers during migration or broad seasonal ranges such as high-use winter ranges. federal government deferred 5674 ha of oil and gas leases after the State of Wyoming designated a mule deer migration corridor to be managed for no net loss of function (Kauffman et al., 2021).
Similar mapping efforts are also informing international conservation efforts. For example, the recently formed Global Initiative on Ungulate Migration, working in partnership with the Convention on Migratory Species, aims to build a digital atlas of ungulate migrations worldwide largely using the approach made available in MM (Kauffman et al., 2021). Such mapping could guide efforts to reduce fencing in the movement corridors of many impacted populations, such as the Loita wildebeest herd in the Mara Ecosystem (Stabach et al., 2022). Detailed migration maps would also advance planning in the Kavango-Zambezi Transfrontier Conservation Area, where connectivity between protected areas is mandated by an international treaty signed by five southern African countries (Kauffman et al., 2021). Finally, development projects such as China's Belt and Road Initiative are occurring where world's migrations still exist (Laurance et al., 2014), and detailed maps can facilitate maintenance of movement corridors amid planned development.
The human footprint continues to grow at an alarming rate, adding additional impediments and challenges to mobile animals globally (Tucker et al., 2018). Within the next 25 years, for instance, humans are expected to build 25 million km of new roads worldwide (Laurance et al., 2014). Successful conservation of wide-ranging species is, in part, dependent on transforming large datasets of animal location data into meaningful polygons and maps that are used to guide planning decisions. MM provides a rigorous yet user-friendly platform to accomplish these tasks. With the data revolution upon us and the increasing size of GPS datasets, MM will help to empower a wider set of biologists to use GPS data to inform local and regional management and conservation.

CO N FLI C T O F I NTE R E S T
Authors declare no conflict of interest.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/2041-210X.13976.

DATA AVA I L A B I L I T Y S TAT E M E N T
Code for Migration Mapper is archived in Zenodo (Merkle et al.,