weatherxbiodiversity-projection

Iberian Bombus extirpation under DestinE Climate DT SSP3-7.0 — replication of Soroye et al. (2020). Reference paper: Soroye et al. (2020)

This repository is a self-contained replication of the headline claim of Soroye et al. (2020), “Climate change contributes to widespread declines among bumble bees across continents”, applied to Iberian Bombus with a future projection under DestinE Climate DT SSP3-7.0.

The replication has two tiers:

• Tier 1 — historical fit. Replicates Soroye’s GLMM on Iberian Bombus using Soroye’s own CRU TS 3.24.01 climate inputs (Figshare deposit) and own-issued GBIF download DOIs. Run twice: once on the original CEA grid (~100 km), once on HEALPix nside=64 (~92 km).

• Tier 2 — future projection. Applies the fitted GLMM to DestinE Climate DT SSP3-7.0 (IFS-NEMO standard, native HEALPix nside=128) for the 2020–2029 and 2030–2039 horizons. The only place new climate data enters.

Headline result — Tier 1¶

Soroye’s central biological claim — that thermal-niche exceedance increases extirpation probability — replicates on Iberian Bombus at both spatial substrates. The GLMM coefficient on sc_TEI_delta is positive, large, and credible:

A higher-resolution sibling study at HEALPix nside=128 (weatherxbiodiversity-projection-nside128) finds +0.347, HDI [+0.139, +0.533] — within ±30% of CEA, so the Tier-1 finding holds across three independent pixelisations.

Headline result — Tier 2¶

Under SSP3-7.0 at 2030–2039, mean future TEI rises from ~0.43 (historical) to ~0.45–0.50, a systematic shift toward each species’ upper thermal niche edge. None of the studied Iberian Bombus species hits TEI > 1 (Soroye’s actual extirpation threshold) at any currently-occupied cell over the next 15 years — the “extirpation event” is a longer-timescale phenomenon, with 2030–2039 being the early-warning period.

The substrate-stable high-risk and low-risk rankings (filtered to species with ≥10 occupied cells per substrate, using main-effects-only η — see methodology in weatherxbiodiversity-substrate-sensitivity):

Biologically coherent: B. terrestris is the dominant Western Palearctic generalist; B. humilis, muscorum, ruderarius are short-tongued grassland specialists already in decline elsewhere in Europe.

Caveat — small-N species cannot be ranked reliably¶

Per-species ranking under SSP3-7.0 is grid-coupled at the projection step for species with fewer than ~10 historical Iberian cells. This affects 18 of 31 species, including the narrowly-distributed Pyrenean specialists (B. pyrenaeus, B. mucidus, B. mendax, B. wurflenii) — precisely the species the public most worries about. The substrate-sensitivity diagnostic is documented in the sibling repo weatherxbiodiversity-substrate-sensitivity.

Quick start¶

git clone https://github.com/annefou/weatherxbiodiversity-projection.git
cd weatherxbiodiversity-projection
mamba env create -f environment.yml
mamba activate weatherxbiodiversity-projection
snakemake --cores 1            # Tier 1 (CEA + HEALPix) — runs anywhere
snakemake --cores 1 tier2      # Tier 2 (DestinE) — needs DestinE credentials

Tier 1 runs end-to-end on a generic machine (Mac, Linux, GitHub Actions). Tier 2 requires DestinE Climate DT credentials (LUMI polytope endpoint) and only runs on the DestinE Jupyter platform or a runner with valid credentials.

Or with Docker (Tier 1 only):

docker run --rm ghcr.io/annefou/weatherxbiodiversity-projection:latest

Pipeline structure¶

The pipeline notebooks are organised in the left-hand TOC into three tiers:

• Tier 1 — CEA replication of Soroye et al. 2020 (notebooks 01–04)

• Tier 1 — HEALPix nside=64 extension (notebooks 02h–04h)

• Tier 2 — DestinE Climate DT SSP3-7.0 projection (notebooks 05–08)

Citation¶

If you use this work, please cite:

• This software: CITATION.cff → DOI Fouilloux (2026).

• The original paper: Soroye et al. (2020).

• The sibling substrate-extension and substrate-sensitivity repos via their own DOIs (see CITATION.cff references).