Tools
Tools
Jung, M. (2023). An integrated species distribution modelling framework for heterogeneous biodiversity data. Ecological Informatics 76 e102127. 10.1016/j.ecoinf.2023.102127.
Preview |
Text
1-s2.0-S1574954123001565-main.pdf - Published Version Available under License Creative Commons Attribution. Download (5MB) | Preview |
Abstract
Most knowledge about species and habitats is in-homogeneously distributed, with biases existing in space, time and taxonomic and functional knowledge. Yet, controversially the total amount of biodiversity data has never been greater. A key challenge is thus how to make effective use of the various sources of biodiversity data in an integrated manner. Particularly for widely used modelling approaches, such as species distribution models (SDMs), the need for integration is urgent, if spatial and temporal predictions are to be accurate enough in addressing global challenges.
Here, I present a modelling framework that brings together several ideas and methodological advances for creating integrated species distribution models (iSDM). The ibis.iSDM R-package is a set of modular convenience functions that allows the integration of different data sources, such as presence-only, community survey, expert ranges or species habitat preferences, in a single model or ensemble of models. Further it supports convenient parameter transformations and tuning, data preparation helpers and allows the creation of spatial-temporal projections and scenarios. Ecological constraints such as projection limits, dispersal, connectivity or adaptability can be added in a modular fashion thus helping to prevent unrealistic estimates of species distribution changes.
The ibis.iSDM R-package makes use of a series of methodological advances and is aimed to be a vehicle for creating more realistic and constrained spatial predictions. Besides providing convenience functions for a range of different statistical models as well as an increasing number of wrappers for mechanistic modules, ibis.iSDM also introduces several innovative concepts such as sequential or weighted integration, or thresholding by prediction uncertainty. The overall framework will be continued to be improved and further functionalities be added.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | Species distribution model; Data integration; Environmental niche; Offset; Point-process-model; Bayesian; R-package |
| Research Programs: | Biodiversity and Natural Resources (BNR) Biodiversity and Natural Resources (BNR) > Biodiversity, Ecology, and Conservation (BEC) |
| Related URLs: | |
| Depositing User: | Michaela Rossini |
| Date Deposited: | 01 Jun 2023 08:46 |
| Last Modified: | 05 Jan 2024 13:47 |
| URI: | https://pure.iiasa.ac.at/18835 |
Actions (login required)
 |
View Item |