Skip to content
Field Notes

The research side of permaculture.studio.

Verification record

Every claim in this article was checked against the abstracts of the papers it cites, by a separate pass that was given those abstracts and nothing else. A claim marked supported carries the sentence that supports it, quoted verbatim below.

References
14
Claims checked
5
Fully supported
4
Checked on
Nov 15, 2026

Claim by claim

  1. A systematic evaluation found that global biodiversity models using study attributes as random effects had low predictive accuracy.

    SupportedMeta-analysisMeta-analysis

    “mixed models with study attributes as random effects, common in biodiversity meta-analysis and indicators, exhibit low predictive accuracy”

    The abstract directly confirms that mixed models using study attributes as random effects exhibit low predictive accuracy in their evaluation of global models.

    Checked by google/gemini-3.8-flash on Nov 15, 2026

  2. Land-use effects estimated by these models were in line with previous research, so the failure is in prediction rather than in the direction of the estimated impact.

    Partially supportedMeta-analysisMeta-analysis

    “despite land-use impacts in line with previous research, there is a challenging gap between effect size inference and local prediction”

    The abstract notes that land-use impacts align with previous research despite a gap in local prediction, but it does not specifically discuss the 'direction' of the estimated impact.

    Checked by google/gemini-3.8-flash on Nov 15, 2026

  3. Replacing study random effects with biome, realm and taxonomic parameters improved predictions, but results remained modest in the contexts sampled.

    SupportedMeta-analysisMeta-analysis

    “Ecologically structured models that replace study random effects with biome, realm and taxonomic parameters show improved but still modest results in sampled contexts.”

    The abstract directly confirms that replacing study random effects with biome, realm, and taxonomic parameters led to improved but modest results in the sampled contexts.

    Checked by google/gemini-3.8-flash on Nov 15, 2026

  4. Predictive performance stayed low when models were extended to other contexts, which the authors relate to shifts in environmental conditions and biodiversity responses.

    SupportedMeta-analysisMeta-analysis

    “performance when extending predictions to other contexts remains low, related to distribution shifts in environmental conditions and biodiversity responses.”

    The abstract directly states that performance remains low when extending predictions to other contexts due to shifts in environmental conditions and biodiversity responses.

    Checked by google/gemini-3.8-flash on Nov 15, 2026

  5. The authors attribute the constraint to current data availability, including geographic and taxonomic biases.

    SupportedMeta-analysisMeta-analysis

    “their applicability is fundamentally constrained by current data availability, which includes geographic and taxonomic biases”

    The abstract directly states that the applicability of the models is fundamentally constrained by current data availability, specifically citing geographic and taxonomic biases.

    Checked by google/gemini-3.8-flash on Nov 15, 2026

What this does and does not prove

It proves that every DOI cited resolves to a real registered work whose returned DOI matches the one requested, that none of them is retracted, and that each claim has a named sentence in a cited abstract behind it.

It does not prove that the paper as a whole supports the claim. An abstract can overstate its own findings, and a sentence can be read out of context. That is why the quote is printed here rather than summarised: if the check is wrong, you can see that it is wrong.

Evidence grading is automatic and conservative. Anything it cannot classify is marked unclassified rather than flattered upward.