Fig 1.
A conceptual diagram illustrating the development of the Invasive Species Habitat Tool (INHABIT).
Co-production starts on the left and follows through to eventual adoption and land management or policy actions. The grey middle circle denotes the iterative cycle where updates are made to species distribution models and user interface elements.
Fig 2.
The layout and features of the Invasive Species Habitat Tool (INHABIT; http://gis.usgs.gov/inhabit) landing page in August 2021.
The sidebar panel is visible on the left and the main panel on the top right with the “Map” tab selected.
Fig 3.
Example content clipped from the “Model Details” tab.
Each chart’s symbology also indicates model type: boosted regression trees (BRT), generalized linear model (GLM), multiadaptive regression splines (MARS), maximum entropy (MAXENT), random forest (RF), and background point generation method (target or kernel density estimate [KDE]). A) Evaluation metrics; in each column, the first number is the value with full model training data and the second, in parentheses, is the mean value from 10-fold cross-validation splits. B) Response curve images representing relative habitat suitability across environmental predictor values. Numbers in the upper left corner within each plot indicate mean, low, and high variable importance values reading from left to right. C) A tabular view of variable importance. Each row displays predictor names. D) Predictor descriptions and hyperlinks to variable sources, and E) the model credibility table based on the model assessment rubric from Tables 1a-c in [8]. Though it is a static template currently based on the overall modeling framework, the future version of this table will update its evaluative criteria relative to models for the selected species.
Fig 4.
Maps of fountain grass habitat suitability within Joshua Tree National Park.
The top map displays the full park extent while the bottom inset map displays the independently surveyed validation points (not included in the Invasive Species Habitat Tool [INHABIT] models) provided by the National Park Service.
Fig 5.
Maps of median normalized difference moisture index (April to September).
This new predictor was reviewed favorably by the U.S. Fish and Wildlife Service for future model iterations due to its ability to delineate water bodies and seasonally inundated wetlands. Two example refuges from Oregon, USA are shown: Bear Lake Wildlife Refuge (left) and Malheur National Wildlife Refuge (right).