Muy BIEN: Next steps in a global workflow for integrating plant botanical observations
Bibliographic record
Abstract
This is my talk given at the 2018 Ecological Society of America meeting in New Orleans, USA. This talk was part of a symposium ' SYMP 9: Ecoinformatics Advances: Building Technosocial Systems for Open Data and Big Science' Wednesday, August 08, 2018 https://eco.confex.com/eco/2018/meetingapp.cgi/Session/14041 https://eco.confex.com/eco/2018/meetingapp.cgi/Paper/70083 Muy BIEN: Next steps in a global workflow for integrating plant botanical observations Brian J. Enquist1, Cory Merow2, Brian J. McGill3, Brad Boyle4, Nathan Casler5, Xiao Feng6, Brian S. Maitner1, Jeanine McGann7, Daniel Park8, Patrick Roehrdanz9 and Lee Hannah10, (1)Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ, (2)Department of Ecology and Evolutionary Biology, Yale University, New Haven, CT, (3)School of Biology and Ecology / Mitchell Center for Sustainability Solutions/Mitchell Center for Sustainability Solutions, University of Maine, Orono, ME, (4)Ecology and Evolutionary Biology Department, University of Arizona, Tucson, AZ, (5)Planet Labs, San Francisco, (6)Ecology and Evolutionary Biology, University of Arizona, AZ, (7)Dept. of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ, (8)Organismic and Evolutionary Biology, Harvard University, Cambridge, MA, (9)Conservation International, Washington DC, DC, (10)Conservation International, Washington DC Background/Question/MethodsOur goal is to develop, and make freely available, a generic pipeline capable of (i) linking biodiversity occurrence data to species ranges and (ii) forecasting and hindcasting those distributions. While such links for one or a few species is now trivial (indeed a few lines of code is all that is needed), scaling these computations to forecast species distributions 1000s or 100,000s of species remains logistically prohibitive for most researchers. A lack of appropriate tools and a failure to combine tools into an integrated pipeline prevent such scaling. Key challenges include: 1) appropriately scrubbing data to remove taxonomic and geographic errors, 2) identifying clear best practice methods for range modelling applicable across diverse species, 3) innovating range modelling methods that integrate diverse data such as presence only museum collections and abundance-based plot data 4) scaling computationally-intensive range modelling in an HPC environment, and 5) placing the outputs of the products in a phylogenetic context, which is increasingly important to conservation efforts. The talk will focus on our progress toward developing such a pipeline using the Botanical Information and Ecology Network (BIEN) , which has assembled a database of 110,000,000 observations of 300,000+ species of plants and the new world. The BIEN project provides a pre-existing user community spanning museum directors, plot ecologists, trait data, and biodiversity scientists.Results/ConclusionsThe BIEN project contains enough data and ecoinformatics tools to demonstrate scalability, and is used to test key assumptions in conservation biology about the phylogenetic conservatism of species climatic niches and the geographic constancy of diversity hotspots over time.The ongoing research is striving to contribute to (i) scientific infrastructure through the development of a scientific codebase for integrating and standardizing heterogeneous sources of observation data in biodiversity science and (ii) to the production of high-quality species ranges from primary biodiversity data. Refinements of the BIEN pipeline will further the development and enable the public release of a massive and freely accessible database compiling occurrence, community, phylogeny, and trait data for all plants. These products are increasingly becoming available online and will benefit both basic and applied research in biodiversity science particularly in conservation and application to citizen science.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.018 | 0.035 |
| Meta-epidemiology (narrow) | 0.004 | 0.002 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.007 | 0.004 |
| Science and technology studies | 0.003 | 0.001 |
| Scholarly communication | 0.009 | 0.013 |
| Open science | 0.006 | 0.015 |
| Research integrity | 0.003 | 0.006 |
| Insufficient payload (model declined to judge) | 0.039 | 0.028 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".