- Lifemapper Data
- Lifemapper Data Terms
Lifemapper data consists of datasets represented by different data objects which contain data and metadata about that data. Analyses can be performed on individual species or taxa, or on large groups of taxa.
The original Lifemapper ‘archive’ contains SDM maps for species with over 30 data points from the Global Biodiversity Information Foundation (GBIF) and is available at http://lifemapper.org. The SDM maps show areas with similar climatic conditions to areas in which the species has been found.
The original Lifemapper archive may be queried to find or summarize data, or to request computations on existing or user-submitted data. Users need not login to query for existing public data, but must register and login to request analyses. Additional installations of the Lifemapper archive focus on regions or species of interest, such as the installation at National Center for High-Performance Computing (NCHC, https://www.nchc.org.tw/en/) in Taiwan.
Lifemapper Data Terms
- All analysis requests are tied to a “gridset”. Gridsets contain one or more species, SDM parameters and scenarios. Optionally, data may be tied to the GBIF Backbone taxonomy, or tied to a provided taxonomy for included species, for subsetting by taxonomy. Other optional elements may be included for multi-species analyses.
Single species data and analyses
Data objects containing geospatial data about individual species include:
- An algorithm is a procedure or formula for solving a problem. There are multiple algorithms for computing Species Distribution Models (SDM) which define the relationship between a set of points and the environmental values at those points. Lifemapper provides 12 algorithms
- Point data representing specimens collected for a single species or taxa. Data contains a location, x and y, in some known geographic spatial reference system. Public data in Lifemapper installations are in the ‘Geographic’ spatial reference system, latitude and longitude in decimal degrees. API documentation is at Occurrence Sets
- Scenarios consist of a set of environmental layers (i.e. elevation, precipitation, temperature, soil, etc). For Species Distribution Modeling, researchers often use inputs of ‘present day’ species points and ‘present day’ climate and/or other environmental data. The environmental data may be global, or regional. The resulting models can be projected onto the same environmental dataset or one predicted for a different time period, or one for another region. An example of predicted environmental data available in the Lifemapper archive is climate data computed for the International Panel on Climate Change (IPCC) for its Fifth Assessment Report (AR5, 2013). API documentation is at Scenarios
- SDM Projections
- Computed SDM models may be applied, or projected back onto the same, or
matching Scenarios. A map created from the projection of this model onto
a Scenario is called a Projection, and is a file of geospatial data in
raster format. Different algorithms produce projections with different values.
The Maxent algorithm produces projections with values denoting the predicted presence as a value between 0 and 1. Other algorithms produce raster files with only the values 1 (predicted present) or 0 (not predicted present).
- The organizing data structure for a single, or group of analyses. A package may be downloaded for visualization within the Lifemapper web client, or further analyses in other software.
Species Distribution Modeling (SDM) experiments follow a general workflow.
- begins with input species data and environmental data
- chooses an algorithm and any parameters for computing the model
- chooses to project the computed model onto the same environmental data, or “matching” data (same type of data and measurement units, perhaps for a different geographic area, or the result of models predicting past or future values)
To request Lifemapper SDM computations, a Lifemapper user may choose inputs from the public data archive, or upload their own data. Species data is held in ‘OccurrenceSet’ objects in the Lifemapper data archive, and may be queried with ‘list’ or ‘count’ queries, using filters as the user
Multi-species data and analyses
Multi-species analysis allow large scale analyses of the distribution of many species. Scale may refer to the taxonomic, phylogenetic, or geographic breadth of the analyses.
Inputs may start with:
- Species layers. These layers can be raster or vector format, and can be predicted (i.e. SDM output projections) or other types of distribution or range maps.
- A grid definition. This defines the geographic bounding box, geographic projection, and grid cell size and shape
- A phylogenetic tree. This contains a tree in NEXUS format with leaves corresponding to each species layer in the analysis.
Biogeographic Hypotheses. These can be in the form of raster or vector files describing biogeographic hypotheses for testing.
- a grid definition and intersection parameters for intersecting SDM projections to produce a Presence-Absence Matrix (PAM).
- a phylogenetic tree, for analyzing the evolutionary patterns in the spatial distribution
- biogeographic hypothesis, for looking at the role of other spatial attributes (i.e. geology, drainage basins, etc) in the biodiversity of an area.
Data products may include the following matrices:
- Matrices representing the intersection of layers with the grid:
- A Presence Absence Matrix, or PAM, which contains the species distributions represented as a binary matrix of 0/1 indicating presence or non-presence in a grid cell.
- Geographic Reference Information Matrix, or GRIM, containing environmental values
- Matrix representing the Biogeographic hypotheses.
- Matrices representing MCPA calculations
- Environmental Correlation
- Environmental P-Value
- Environmental R-Squared
- Environmental R-Squared P-Value
- Biogeographic Environmental Correlation
- Biogeographic Environmental P-Value
- Biogeographic Environmental R-Squared
- Biogeographic Environmental R-Squared P-Value