Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› conbio-data-analysis
conbio-data-analysis
GitHub指导保护生物学论文的数据分析与报告，确保通过严格的双盲同行评审。涵盖模型选择、不确定性报告、稳健性检验及可重复性规范，避免常见错误，强调结果对保护决策的意义。
Trigger Scenarios
执行主分析和补充分析并撰写结果部分
审稿人要求提供稳健性或替代模型分析
区分探索性与确认性分析
在提交前确保分析过程可复现
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill conbio-data-analysis -g -y
SKILL.md
Frontmatter
{
"name": "conbio-data-analysis",
"description": "Use when executing and reporting the analysis for a Conservation Biology manuscript so it survives expert, double-blind review — appropriate ecological\/statistical models, honest uncertainty, robustness, and reproducibility. Covers detection, hierarchical models, spatial structure, and effect sizes that matter for conservation. Guides analysis norms; it does not fabricate results."
}
Data Analysis (conbio-data-analysis)
Conservation Biology reviewers are methodologically sophisticated, and the journal expects a
data-availability statement with data and code deposited at acceptance (see
conbio-reporting-and-data-policy). Analyze as if your code will be re-run — because it may be. This
skill covers execution and reporting norms; design decisions live in conbio-study-design.
When to trigger
- Running main and supporting analyses; building the results section
- A reviewer asked for robustness, alternative models, or uncertainty
- Reconciling exploratory vs. confirmatory analyses
- Making the analysis reproducible before deposit
Analysis norms Conservation Biology expects
- Report uncertainty honestly. Confidence/credible intervals, not just stars or p-values; report the magnitude and conservation meaning of the estimate, not only significance.
- Use the right model for the data. Hierarchical/mixed models for nested data; occupancy and N-mixture for detection; capture-recapture for survival/abundance; GLMs/GAMs for nonlinearity; account for spatial autocorrelation and zero-inflation where present.
- Robustness that probes, not decorates. Show specifications that could break the result (alternative predictors, samples, priors, estimators), and say what you learn.
- Right inference. Cluster/group at the correct level; avoid pseudoreplication in the analysis; correct for multiple comparisons when testing many implications.
- Confirmatory vs. exploratory. Separate preregistered/confirmatory tests from exploratory ones; do not mine for a significant interaction and theorize it post hoc.
- Model checking. Report convergence, residual diagnostics, validation/out-of-sample performance for predictive models; show the result is not an artifact of one modeling choice.
Conservation-specific reporting
- Translate estimates into decision-relevant quantities (extinction risk, population trend, effect of a management action, area needed) with uncertainty.
- For projections (PVA, SDM, climate), state the assumptions and the range of plausible outcomes — not a single point forecast.
Reproducibility while you work (not at the end)
- One master script regenerates every table and figure from the (raw or constructed) data.
- Set and report seeds for bootstrap, MCMC, simulation, and any stochastic step.
- Pin software/package versions (
renv.lock,requirements.txt, recorded installs). - Keep table/figure numbers matched to script outputs.
Anti-patterns
- Stars/p-values with no effect sizes or intervals
- Raw counts analyzed as abundance with detection ignored
- "Robustness" that only reruns near-identical specs to manufacture stability
- p-hacking / HARKing exploratory results into confirmatory claims
- A single point projection presented as certain
- A results section whose numbers the code cannot reproduce
Evidence pass for Conservation Biology
Use this as a second-pass capability check. First lock the species/system threat, conservation decision, and uncertainty relevant to action; then test whether the manuscript addresses conservation-science reviewers who ask whether evidence changes biodiversity, management, or policy action.
- Primary move: Audit unit, comparison, uncertainty, missingness, sensitivity, and reproducibility before making any prose or submission recommendation.
- Decision ledger: return
claim / evidence / blocker / next editrows so the next pass can patch the manuscript directly. - Neighbor test: compare against Biological Conservation for applied conservation breadth, Global Change Biology for climate/ecosystem process, Ecology Letters for theory-forward ecology; if the neighboring outlet has the stronger audience claim, recommend re-routing before polishing.
- Submission-ready gate: before final advice, re-open
resources/official-source-map.mdfor upload-week rules and name the one live-check item that could change the recommendation.
Output format
【Main estimate】magnitude + interval + conservation meaning
【Model】why this model fits the data (detection / hierarchy / spatial)
【Robustness】specs that could break it → what held
【Confirmatory vs exploratory】clearly separated?
【Uncertainty in projections】range stated, not a point?
【Reproducible】master script + seeds + pinned versions? [Y/N]
【Next】conbio-figures-and-tables
Supplementary resources
../../resources/external_tools.md— modeling, inference, and synthesis packages../../resources/official-source-map.md— data-availability and reproducibility expectations
Version History
- 1839142 Current 2026-07-05 12:47
