|
| Aim: The geographic range and ecological niche of species are widely used concepts
in ecology, evolution and conservation and many modelling approaches have been
developed to quantify each. Niche and distribution modelling methods require a lit-
any of design choices; differences among subdisciplines have created communication
barriers that increase isolation of scientific advances. As a result, understanding and
reproducing the work of others is difficult, if not impossible. It is often challenging to
evaluate whether a model has been built appropriately for its intended application or
subsequent reuse. Here, we propose a standardized model metadata framework that
enables researchers to understand and evaluate modelling decisions while making
models fully citable and reproducible. Such reproducibility is critical for both scien-
tific and policy reports, while international standardization enables better compari-
son between different scenarios and research groups.
Innovation: Range modelling metadata (RMMS) address three challenges: they
(a) are designed for convenience to encourage use, (b) accommodate a wide variety
of applications, and (c) are extensible to allow the research community to steer them
as needed. RMMS are based on a metadata dictionary that specifies a hierarchical
structure to catalogue different aspects of the range modelling process. The diction-
ary balances a constrained, minimalist vocabulary to improve standardization with
flexibility for users to modify and extend. To facilitate use, we have developed an R
package, rangeModelMetaData, to build templates, automatically fill values from
common modelling objects, check for inconsistencies with standards, and suggest
values.
Main conclusions: Range Modelling Metadata tools foster cross-disciplinary ad-
vances in biogeography, conservation and allied disciplines by improving evaluation,
model sharing, model searching, comparisons and reproducibility among studies. Our
initially proposed standards here are designed to be modified and extended to evolve
with research trends and needs. | |
|