AI evidence extraction

Rapid evidence mapping of soil fauna responses to agricultural management assisted by large language models

Ricardo João Cerdeira Veiga Leitao, Vid Podpečan, Luís Cunha, Carmen Vazquez, Felix David, Camille Imbert et al. · Geoderma · 2026

AI-generated evidence extraction, verified across multiple analytical personas. Not a substitute for the peer-reviewed original.

9/10

Relevance

0/4

Quality (LMQS)

Evidence

Citations

0.00

FWCI

This is an AI analysis. Read the peer-reviewed original at the publisher: https://doi.org/10.1016/j.geoderma.2026.117890

Methodology & findings

Study design

Multi-module LLM-assisted knowledge-extraction workflow using iteratively refined prompt chains applied to scientific abstracts.

Main result

The workflow indicated "predominantly beneficial reported patterns for crop residue retention, particularly for earthworms and nematodes, whereas biochar showed a more heterogeneous and context-dependent pattern." Benchmarking against manually curated datasets showed high precision and recall, demonstrating that LLM-assisted workflows can support rapid, large-scale evidence mapping of soil fauna responses to management practices.

Research paradigm

Empirical-computational (technology-assisted evidence synthesis)

Author conclusions

"Overall, results show that LLM-assisted workflows can support rapid, large-scale evidence mapping of soil fauna responses to management practices when formal quantitative syntheses are unavailable." The authors emphasize that "The complete workflow is publicly available and broadly transferable across environmental research domains."

Risk of bias

Abstract-level extraction may introduce selection bias toward studies with detailed abstracts; LLM hallucination and prompt engineering effects; Potential publication bias in underlying literature; Limited to peer-reviewed literature captured by the underlying meta-data-analysis framework; Limitation to abstract-level information may introduce selection and information bias; Reliance on LLM-extracted data from unstructured text may introduce extraction errors; Manual curation bias in benchmark dataset creation; Potential language bias if literature predominantly in English

Limitations

The authors note that "the information content of abstracts was likely a major practical constraint on extraction performance." They also state that "The proposed framework is best understood as a complementary tool for organising and screening dispersed ecological evidence, and not as a substitute for full-text synthesis, effect-size-based meta-analysis, or decision-grade inference."

Open questions raised

Gap in rapid evidence synthesis methods for soil fauna responses to agricultural management; knowledge gaps specifically identified for biochar and crop-residue retention effects on soil fauna; need for complementary tools to organize dispersed ecological evidence when formal quantitative meta-analysis is unavailable.
The paper identifies a knowledge gap application regarding biochar and crop-residue retention effects on soil fauna, addressing areas where quantitative syntheses are unavailable.

Data: not_statedCode: The complete workflow is stated to be "publicly available" but specific repository location not provided in abstractExtracted from: pdfAgreement 58%

Explore related topics

Systematic Review Automation Literature Discovery Empirical Study